Poly(Homopiperazine-Amide) Thin-Film Composite Membrane for Nanofiltration of Heavy Metal Ions.

The development of membrane-based technologies for the treatment of wastewater streams and resources containing heavy metal ions is in high demand. Among various technologies, nanofiltration (NF) membranes are attractive choices, and the continuous development of novel materials to improve the state-of-the-art NF membranes is highly desired. Here, we report on the synthesis of poly(homopiperazine-amide) thin-film composite (HTFC)-NF membranes, using homopiperazine (HP) as a monomer. The surface charge, hydrophilicity, morphology, cross-linking density, water permeation, solute rejection, and antifouling properties of the fabricated NF membranes were evaluated. The fabricated HTFC NF membranes demonstrated water permeability of 7.0 ± 0.3 L/(m2 h bar) and rejected Na2SO4, MgSO4, and NaCl with rejection values of 97.0 ± 0.6, 97.4 ± 0.5, and 23.3 ± 0.6%, respectively. The membranes exhibit high rejection values of 98.1 ± 0.3 and 96.3 ± 0.4% for Pb2+ and Cd2+ ions, respectively. The fouling experiment with humic acid followed by cross-flow washing of the membranes indicates that a flux recovery ratio (FRR) of 96.9 ± 0.4% can be obtained.

Introduction

The deterioration and contamination of surface and groundpan class="Chemical">water resources by heavy metal ions mandate the need for advanced remediation technologies.[1,2] Due to the hazardous impact of heavy metal ions on human health, complete removal of these ions from water resources is of critical importance.

The term “heavy metals” refers to elements with a specific gpan class="Chemical">ravity greater than 5.0 and atomic weight between 63.5 and 200.6 amu.[3] Among various heavy metal contaminants, lead and cadmium are frequently detected in industrial wastewater.[4] The current manufacturing schemes for car batteries, paints, fertilizers, pigments, etc., are among the main sources of these contaminants in water supplies. The overexposure of humans and animals to these metal ions can cause severe health problems.[5] According to the World Health Organization, the maximum contaminant level (MCL) of lead in drinking water is 15 ppb. It is reported that the presence of 10 ppb of lead in drinking water could cause cognitive impairment.[6] Furthermore, the intake of drinking water contaminated with lead above MCL is reported to be damaging to the kidneys, liver, and both nervous and reproductive systems. Similarly, there are serious concerns about the potential carcinogenic effect of cadmium on human health.[7] Consequently, high demand exists for the development of schemes that allow for reducing the concentration of heavy metal ions in water resources. Chemical precipitation, ion exchange, flotation, and electrochemical treatment are the conventional methods for treating water resources contaminated with heavy metal ions.[7−9] However, all of these methods have their limitations. For instance, the chemical precipitation method is associated with the production of a large quantity of sludge and toxic fumes; it is suitable only for wastewater containing a high concentration of heavy metals.[10] The low efficacy (∼60–90%), the high cost of resin, and the difficulties involved with the regeneration of used resin make the ion-exchange process less viable for heavy metal removal.[11]

Currently, membrane-bpan class="Chemical">ased purification is the most promising and scalable approach for the removal of heavy metal ions from contaminated water resources. Among all of the membrane-based techniques, reverse osmosis (RO) and nanofiltration (NF) are considered as the state-of-the-art technologies for water purification and desalination. NF plays a vital role in wastewater purification and desalination because it can be considered as the intermediate stage between RO and ultrafiltration (UF).[12] Compared to RO, NF operates at lower pressures and exhibits complete rejection of the solute in the range of 100–1000 Da.[13]

Recently, applications of NF have been extended from desalination to the removal of heavy metal ions,[14] dyes,[15] pharmaceutical waste,[16] and pesticides from impaired water resources.[17] The state-of-the-art NF membranes are thin-film composite (TFC) membranes, prepared via interfacial polymerization (IP). A TFC membrane consists of an active polyamide (PA) layer, in situ polymerized on a UF or microfiltration (MF) membrane. The TFC performance in NF primarily depends on the quality of the PA layer, which controls the permeability and solute rejection of the membrane. The properties of the PA layer can be adjusted by the proper choice of support,[18] monomer[19] miscibility of phases,[20] reaction time,[21] temperature,[22] and post-treatment.[23] Among these parameters, the selection of monomer and post-treatment is considered as a facile strategy to improve the TFC NF membrane.[24]

In the fabrication of TFC NF membpan class="Chemical">ranes for heavy metal removal, monomers such as m-phenylenediamine (MPD),[25] polyethylenimine (PEI),[26] piperazine (PIP),[27] chitosan,[28] polydopamine,[29] poly(2-methacryloyloxyethyl phosphorylcholine-co-2-aminoethyl methacrylate),[30] and poly(amidoamine) (PAMAM)[25] and post-treatment using PEI[27] are reported. However, the preparation of the TFC NF membrane with improved water permeability without compromise in solute rejection is highly challenging.

Kumano and Matsuyama reported the fabrication of the outer selective TFC NF hollow fiber membrane using homopiperazine (HP).[31] Nevertheless, the as-prepared membrane exhibited reduced water flux, and no data reported on the heavy metal removal and antifouling property of the TFC membrane. In another study, the chlorine resistance of the TFC membrane prepared by aliphatic and aromatic amines was evaluated.[32] It was observed that the TFC membrane prepared with HP exhibited improved chlorine resistance than MPD. The increased chlorine resistance was attributed to the reduced basicity of HP compared to MPD. Furthermore, the preparation of TFC and thin-film nanocomposite (TFN) membranes using HP as an amine source is also patented for NaCl and Na2SO4 removal and organic solvent nanofiltration.[33−37] However, no studies were performed to evaluate the heavy metal removal and antifouling property of the TFC membranes prepared using HP as a monomer.

In this study, we report on the synthesis of poly(homopiperazine–pan class="Chemical">amide) TFC (HTFC) nanofiltration (NF) membranes by reacting homopiperazine (HP) and trimesoyl chloride (TMC) at the aqueous–organic interface. The as-prepared HTFC NF membranes were further post-treated with ethylenediamine (EDA) in isopropyl alcohol (IPA) to fine-tune the HTFC NF membrane surface properties. To the best of our knowledge, heavy metal ion removal and antifouling properties of post-treated poly(homopiperazine–amide) TFC nanofiltration membranes have not been studied yet. The effect of post-treatment time on the NF membrane permeability, surface roughness, hydrophilicity, and heavy metal ion rejection was examined. Also, the antifouling property of the HTFC NF membranes was characterized using humic acid (HA) as a model foulant in the cross-flow filtration mode.

Results and Discussion

Chemical Characterization of HTFC Membranes

To confirm the chemistry of HTFC membranes, we characterized the membranes with attenuated total reflectance-Fourier transform infrared spectroscopy (ATR-FTIR). Figure A presents the ATR-FTIR spectra of the PSF substrate and the HTFC membranes. All of the HTFC membranes exhibited peaks at 2967, 1293, 1242, and 1150 cm–1, attributed to the characteristic aromatic C–H stretch, S=O asymmetric stretch, C–O–C stretch, and S=O symmetric stretch of the PSF support, respectively.[38] Comparing the spectra of PSF and HTFC membranes, a new peak located at ∼1625 cm–1 in the FTIR spectra of all of the HTFC samples was observed. This peak was ascribed to the C=O stretching vibration of the amide group (amide I).[39] This observation confirms the presence of the PA layer on the PSF substrate. The broad peaks located at ∼3401 and 1725 cm–1 were ascribed to the −OH and C=O stretches of the residual carboxylic acid groups, respectively.

Figure 1

(A) ATR-FTIR spectra of (a) PSF substrate, (b) control, (c) HTFC-IPA, (d) HTFC-1, (e) HTFC-2, and (f) HTFC-3 nanofiltration membranes. (B) X-ray photoelectron spectroscopy (XPS) survey spectra of HTFC membranes. The curve of the control (black) represents the membrane that was not post-treated with ethylenediamine (EDA). The HTFC-IPA (red), HTFC-1 (blue), HTFC-2 (green), and HTFC-3 (orange) represent the spectra for the membranes that were post-treated with EDA.

(A) ATR-FTIR spectpan class="Chemical">ra of (a) PSF substrate, (b) control, (c) HTFC-IPA, (d) HTFC-1, (e) HTFC-2, and (f) HTFC-3 nanofiltration membranes. (B) X-ray photoelectron spectroscopy (XPS) survey spectra of HTFC membranes. The curve of the control (black) represents the membrane that was not post-treated with ethylenediamine (EDA). The HTFC-IPA (red), HTFC-1 (blue), HTFC-2 (green), and HTFC-3 (orange) represent the spectra for the membranes that were post-treated with EDA.

We performed XPS measurements to analyze the elemental composition and cross-linking density of the formed PA layer on the PSF support. Figure B shows the survey spectra of the control and fabricated HTFC membranes. The main features of the spectra are the sharp peaks assigned to C 1s, N 1s, and O 1s. The small peak around 497 eV is associated with Na KLL from sodium hydroxide, which was used to neutralize the as-formed HCl during IP. The high-resolution C 1s, N 1s, and O 1s core electron spectra of the sample before (HTFC-IPA) and after (HTFC-3) post-treatment can be found in Figure ; the XPS spectra for all samples can be found in the Supporting Information. The C 1s core electron spectrum of all samples shows four peaks centered at 284.3, 285.3, 287.2, and 291.2 eV ascribed to C–C, C–N, C=O, and shake-up features, respectively.[40] Three distinct peaks also were identified in N 1s core electron spectra of different samples, centered at 398.2, 399.2, and 400 eV. These peaks were attributed to CNH/CNH2, N–C, and N–C=O species, respectively.[41] The O 1s core electron spectra show two distinct peaks centered at 530.8 and 532.3 eV. These peaks are assigned to C=O and O–H species, respectively.

Figure 2

High-resolution C 1s, N 1s, and O 1s XPS core electron spectra for HTFC nanofiltration membranes before (HTFC-IPA) and after (HTFC-3) post-treatment with EDA.

High-resolution C 1s, N 1s, and O 1s XPS core electron spectra for Hpan class="Chemical">TFC nanofiltration membranes before (HTFC-IPA) and after (HTFC-3) post-treatment with EDA.

Elemental analysis was used to find the degree of cross-linking for samples before and after pan class="Chemical">EDA treatment. For the untreated samples (control and HTFC-IPA), the degree of cross-linking (X) was directly correlated to the O/N ratio;[42] for the detailed calculations, see Figure S2, Supporting Information. Table includes the surface elemental composition (atomic %), O/N ratios, as well as the estimated degree of cross-linking for samples before and after EDA treatment. The O/N ratios for the control and HTFC-IPA membranes were about 1.0 and 1.24, respectively. Thus, by correlating the O/N to the cross-linking density, the control membrane exhibited a degree of cross-linking of 58.3%. However, this value fell to 40.3% after the sample was treated with IPA (HTFC-IPA). This reduction in the degree of cross-linking was attributed to the removal of unreacted monomers and hydrolysis of the acid chloride group in the PA layer by the water-miscible solvent IPA, which reduced further cross-linking during heat treatment.[43]

Table 1

XPS Elemental Composition of HTFC Membranes

	atomic concentration (%)
membrane	C 1s	O 1s	N 1s	Na 1s	Cl 2p	O/N ratio	X (%)	XEDA (%)
control	72.72 ± 0.03	13.4 ± 0.02	13.43 ± 0.05	0.44 ± 0.02		1.0	58.3
HTFC -IPA	73.71 ± 0.02	13.73 ± 0.05	11.08 ± 0.03	1.16 ± 0.03	0.31 ± 0.03	1.24	40.3
HTFC-1	70.74 ± 0.03	14.99 ± 0.03	11.98 ± 0.02	2.01 ± 0.03	0.28 ± 0.02	1.25		23.7
HTFC-2	72.21 ± 0.04	13.52 ± 0.05	12.07 ± 0.03	1.84 ± 0.05	0.35 ± 0.02	1.12		24.1
HTFC-3	72.61 ± 0.02	13.14 ± 0.04	13.67 ± 0.02	0.25 ± 0.05	0.33 ± 0.02	0.96		40.3

From the XPS elemental analysis, it can also be deduced that the O/N pan class="Chemical">ratio decreases from 1.25 (HTFC-1) to 0.96 (HTFC-3). The slight reduction of the O/N ratio after post-treatment (cross-linking using EDA) is due to the additional nitrogen atoms covalently attached to the PA network. To define the degree of cross-linking for the samples that went through EDA treatment, we estimated the concentration of HN–C species using the chemical state and elemental analysis data provided in Tables S1–S3, Supporting Information. To do so, we looked at the C 1s and O 1s spectra simultaneously. First, we subtracted the concentration of O–H groups from the concentration of C=O species in the O 1s spectrum. Here, according to the proposed chemical structure after EDA treatment (Scheme ), the C=O species are connected to either N atoms present in the HP rings (O=C−N) or NH groups of the linkers (O=C−NH). If we multiply the resultant value by two, it will be approximately equal to the total concentration of C–N groups located in the HP rings. Subsequently, by subtracting the concentration of C–N groups located in the HP rings from the total concentration of C–N in the C 1s spectrum, we found the equivalent concentration of C–NH species in the linkers. We defined the cross-linking density of the treated samples (XEDA) to be the ratio of C–NH species to the total C–N concentration, as shown in eq S1, Supporting Information. As shown in Table , the XEDA value for HTFC-1 was about 23.7%, and it increased to about 24.1 and 40.3% for the HTFC-2 and HTFC-3 samples, respectively. This enhancement in the degree of cross-linking shows the effectiveness and evolution of EDA treatment with reaction time.

Scheme 1

Synthetic Scheme of Poly(Homopiperazine–Amide) Thin-Film Composite (HTFC) Membrane Preparation

In the first step, homopiperazine (HP) and trimesoyl chloride (TMC) reacted via the Schotten–Baumann reaction to give polyamide-I. In the second step, ethylenediamine (EDA) was post-treated with the as-formed polyamide-I layer to react with the residual acyl chloride group and yield polyamide-II.

In the first step, homopiperazine (pan class="Chemical">HP) and trimesoyl chloride (TMC) reacted via the Schotten–Baumann reaction to give polyamide-I. In the second step, ethylenediamine (EDA) was post-treated with the as-formed polyamide-I layer to react with the residual acyl chloride group and yield polyamide-II.

Another critical property of the PA layer is the surface morphology of the as-prepared HTFC membranes. For this reason, the morphology of membranes was characterized using scanning electron microscopy (SEM) and atomic force microscopy (AFM) techniques. The cross-section SEM images of the control and an HTFC membrane are shown in Figure S3A, where the thickness of the PA layer was measured to be ∼80 nm. Figure A shows the formed nodular and globular structures on the surface of the control and HTFC-IPA membranes. The AFM height images of the control and HTFC-IPA membranes, shown in Figure B, confirm the presence of a globular structure on the top layer. The estimated average roughness (Ra) factor of the control membrane surface is about 36% higher than that of the HTFC-IPA sample. The difference in Ra was ascribed to the molecular rearrangement induced by treating the membrane with an organic solvent, resulting in irreversible structural deformation of the PA layer.[43,44] We did not note any trend in the roughness factor of the membranes as a function of post-treatment processing. Nonetheless, our HTFC membranes show lower roughness compared with that shown in most published works.[45,46] The top-surface SEM and AFM images of HTFC-1, HTFC-2, and HTFC-3 can be found in Figure S3B.

Figure 3

(A) Top-surface SEM and (B) AFM two-dimensional (2D) images of the control and HTFC-IPA NF membranes. The control membrane was not washed with IPA, whereas the HTFC-IPA membrane was washed with 20 mL of IPA for 1 minute at 20 °C. Both SEM and AFM images depict the presence of nodular and globular structures of control and HTFC-IPA membranes.

(A) Top-surface SEM and (B) AFM two-dimensional (2D) images of the control and HTFC-IPA NF membpan class="Chemical">ranes. The control membrane was not washed with IPA, whereas the HTFC-IPA membrane was washed with 20 mL of IPA for 1 minute at 20 °C. Both SEM and AFM images depict the presence of nodular and globular structures of control and HTFC-IPA membranes.

Figure A presents the ζ-potential for all of the HTFC membpan class="Chemical">ranes in the pH range of 5–9. As shown, all of the as-prepared HTFC membranes demonstrated a positive charge below pH 6.2, which favors the rejection of divalent or multivalent ions via the Donnan effect.[47] The isoelectric point (IEP) of the control membrane was found to be at pH 6.47. Compared with the control samples, the IEP for the HTFC-1 membrane declines to pH 6.25. The reduced IEP (slightly increased negative charge) was attributed to the removal of unreacted monomers and oligomer during the post-treatment process. Because of the unreacted amine removed during the post-treatment, the unreacted acyl group (−COCl) in the PA layer hydrolyzed in the presence of water, forming a COOH group. As a result, the HTFC-1 membrane becomes more negative than the control membrane. By increasing the post-treatment time, we expect the unreacted carboxylic groups to be cross-linked by EDA, which results in an increased IEP.

Figure 4

(A) ζ-Potential as a function of pH. ζ-Potential analysis was performed using 1 mM KCl as the background electrolyte. For each measurement, the pH of the electrolyte was adjusted by an auto titrator using 0.05 M NaOH or 0.05 M HCl solutions. The curve of the control (black) represents the membrane that was not post-treated. The curve of HTFC-1 (red) represents the membrane that was post-treated with EDA in IPA for 1 min. The curves of HTFC-2 (green) and HTFC-3 (blue) depict the membranes that were post-treated with EDA in IPA for 2 and 3 min, respectively. (B) Contact angle of all of the HTFC NF membranes. The measurements were performed at three different locations on each sample. The reported values show the average of three measurements with one standard deviation.

(A) ζ-Potential as a function of pH. ζ-Potential analysis wpan class="Chemical">as performed using 1 mM KCl as the background electrolyte. For each measurement, the pH of the electrolyte was adjusted by an auto titrator using 0.05 M NaOH or 0.05 M HCl solutions. The curve of the control (black) represents the membrane that was not post-treated. The curve of HTFC-1 (red) represents the membrane that was post-treated with EDA in IPA for 1 min. The curves of HTFC-2 (green) and HTFC-3 (blue) depict the membranes that were post-treated with EDA in IPA for 2 and 3 min, respectively. (B) Contact angle of all of the HTFC NF membranes. The measurements were performed at three different locations on each sample. The reported values show the average of three measurements with one standard deviation.

Another way to chapan class="Chemical">racterize the polarizability of the surface is to evaluate the wettability of the surfaces by measuring the water contact angle (CA).[46] As shown in Figure B, the control membrane has the lowest water CA of 16.9° due to the higher Ra value when compared to all other HTFC membranes. According to the Wenzel model, the increased surface roughness will reduce the contact angle.[48] Here, an increase in the contact angle was noted as a function of the treatment process, with the HTFC-3 membrane demonstrating the highest contact angle of all surfaces. This trend in the CA values was attributed to the increased cross-linking density and the reduced availability of free carboxylic acid groups in the PA layer.[49]

HTFC Membrane Performance

The water permeability and effective solute rejection of the HTFC membranes can be tuned by adjusting the cross-linking density of the PA layer and post-treatment.[43,50] Often, increasing the cross-linking density results in the formation of a tighter polymeric network, which causes enhanced solute rejection and reduced water permeability. Here, we showed that post-treatment with organic solvent and amine group results in creating an NF membrane with a reduced cross-linking density and a positively charged surface. The aliphatic nature of the used diamine and the organic solvent allows the NF to swell and cross-link with an extended degree of motion compared to the cross-linking of the rigid aromatic precursors. Thus, the NF membrane prepared by diamine cross-linking is expected to have a more open structure and a positive charge, as evidenced by the IEP measurement.

To estimate the nominal pore size and the solute rejection efficiency of the PA layer, we evaluated the rejection values for different molecular weight PEGs; the MWCO of all of the membranes was also estimated. As shown in Figure A, the MWCO values of the control, HTFC-IPA, HTFC-1, HTFC-2m, and HTFC-3 membranes were 253, 260, 272, 266, and 326 D, respectively. The corresponding Stokes radius (rp) values for these membranes were estimated to be 0.36, 0.37, 0.38, 0.37, and 0.42 nm. The MWCO values of all of the NF membranes were in the same range except for the HTFC-3 membrane. The slight decrease in PEG rejection was ascribed to the extended post-treatment processing time. We postulate that during post-treatment, not only the amidation reaction between EDA and TMC happens, but also the removal of the low-molecular-weight PA layers occurs. Consequently, structural defects are prone to be formed. The latter could be the reason leading to a slight reduction in PEG rejection, increasing the MWCO of the HTFC-3 membrane.

Figure 5

(A) Molecular weight cutoff (MWCO) analysis of the control and HTFC NF membranes. MWCO values of all of the membranes were determined by filtering 200 ppm aqueous PEG solution at 150 psi (10.3 bar) and 20 °C; (B) 2000 ppm Na2SO4, MgSO4, and NaCl salt solution flux (L/(m2 h)) of NF membranes at 150 psi (10.3 bar) and 20 °C; (C) 2000 ppm Na2SO4, MgSO4, and NaCl salt rejection of NF membranes at 150 psi (10.3 bar) and 20 °C. (D) Long-time Na2SO4 (2000 ppm) rejection and salt solution flux of HTFC-1 membranes at 150 psi and 20 °C (the curve (black) with square-shaped dots indicates the flux, and the curve (red) with round-shaped dots indicates the rejection).

(A) Molecular weight cutoff (MWCO) analysis of the control and HTFC NF membpan class="Chemical">ranes. MWCO values of all of the membranes were determined by filtering 200 ppm aqueous PEG solution at 150 psi (10.3 bar) and 20 °C; (B) 2000 ppm Na2SO4, MgSO4, and NaCl salt solution flux (L/(m2 h)) of NF membranes at 150 psi (10.3 bar) and 20 °C; (C) 2000 ppm Na2SO4, MgSO4, and NaCl salt rejection of NF membranes at 150 psi (10.3 bar) and 20 °C. (D) Long-time Na2SO4 (2000 ppm) rejection and salt solution flux of HTFC-1 membranes at 150 psi and 20 °C (the curve (black) with square-shaped dots indicates the flux, and the curve (red) with round-shaped dots indicates the rejection).

We also measured the pan class="Chemical">water permeability, A, of all of the HTFC membranes. The data are presented in Figure S4. By comparing the pristine control membrane with the HTFC-1 NF membrane, an increase in the A parameters from 3.6 ± 0.4 to 7.0 ± 0.3 L/(m2 h bar) can be noted. This increase in water permeability was attributed to the post-treatment process. Additionally, the pristine membranes demonstrated a higher cross-linking density, as reflected in Table , which can contribute to the inferior permeability of these membranes. When the EDA solution post-treatment time was increased, water permeability decreased. This change was ascribed to the increased cross-linking density of the active layer. EDA molecules in IPA reacted with unreacted acid chloride and resulted in a higher cross-linking density. These results are in agreement with the literature, indicating that the permeability of the NF membrane mainly depends on the porosity of the PA layer,[51] surface hydrophilicity,[52] and PA-layer thickness.[53]

To evaluate the performance of the membranes in rejecting the solute, we chose divalent and monovalent salts such as Na2SO4, MgSO4, and NaCl and prepared feed solutions with 2000 ppm concentration of these salts. The rejection and permeation data for the NF membranes are shown in Figure B,C. All of the NF membranes demonstrated more than 94% rejection toward Na2SO4 and MgSO4. Subsequently, all of the HTFC NF membranes exhibited below 30% rejection for NaCl. The improved rejection of MgSO4 was attributed to the Donnan exclusion effect. The control and HTFC-1 membranes were negatively charged, while HTFC-2 and HTFC-3 membranes were positively charged at pH 7 (Figure A). Therefore, in the cases of the control and HTFC-1 membranes, the SO42– ions were electrostatically repelled. Also, Mg2+ ions were rejected to maintain electrical neutrality. On the other hand, for the HTFC-2 and HTFC-3, Mg2+ ions were rejected via electrostatic repulsion. Additionally, the Stokes radius values for the HTFC-2 and HTFC-3 membranes were estimated to be 0.37 and 0.42 nm, respectively, while the hydrated ionic radius values of SO42–, Mg2+, Cl–, and Na+ ions are reported to be 0.4, 0.43, 0.33, and 0.36 nm, respectively.[54,55] As a result, the Na2SO4 rejection mechanism for HTFC-2 and HTFC-3 membranes depends on size exclusion. The obtained results are well aligned with the literature.[56]

From a practical point of view, opepan class="Chemical">rational stability is one of the main requirements for the NF membrane. For this reason, an HTFC-1 membrane was chosen to be characterized for long-term tests (24 h) due to better permeability and salt rejection. As shown in Figure D, the HTFC-1 membrane demonstrated promising stability, permeability, and salt rejection over 24 h. The obtained results indicated that the as-prepared HTFC membranes are suitable for desalination.

Heavy Metal Ion Removal and Antifouling Study

To explore the versatility of the as-prepared HTFC membranes, we tested the heavy metal ion rejection efficiency of all of the HTFC NF membranes, presented in Figure A. All of the HTFC membranes exhibited >97% Pb2+ and >94% Cd2+ ion rejection at pH 5. As Pb2+ and Cd2+ ions form insoluble metal hydroxides at above pH 7,[57] in this study, the rejection experiment was performed at pH 5. In a review of the metal ion rejection results, the HTFC NF membranes demonstrated higher rejection toward Pb2+ than Cd2+ ions. This is unexpected, as Pb2+ ions are smaller than Cd2+ ions. The reason for the higher rejection of Pb2+ over Cd2+ can be explained as follows: (i) higher normalized volume charge density and (ii) lower ionic strength of the Pb(NO3)2 solution than the Cd(NO3)2 solution and (iii) increased hydrated stability of Pb2+ at pH 5 than Cd2+. The increased hydrated stability of Pb2+ bestowed better charge–charge repulsion between the Pb2+ ions and the positively charged membrane surface.[27] Therefore, Pb2+ ions were rejected by >97% when compared to the Cd2+ ions (>94%). The heavy metal ion rejection ability of the as-prepared membrane was compared with the literature, and the results were comparable to most of the membranes and superior to some of the NF membranes reported (Table ).

Figure 6

(A) Pb2+ and Cd2+ rejection of the HTFC-1 membrane at pH 5, 150 psi (10.3 bar), and 20 °C. The filtration was performed with 10 ppm of Pb(NO3)2 and Cd(NO3)2 aqueous solution individually. (B) Antifouling performance of the control, HTFC-IPA, HTFC-1, HTFC-2, and HTFC-3 membranes with 200 ppm of aqueous humic acid (HA) as feed at 150 psi (10.3 bar) and 20 °C. Water permeability was measured in the first 8 h using DI water as a feed solution. Then, the feed solution was replaced with 200 ppm aqueous HA, and filtration was continued for another 8 h. Finally, the membranes were washed with DI water, and again the water permeability was measured for another 8 h using DI water as feed.

Table 2

Comparison of Heavy Metal Removal Capacity of As-Prepared NF Membranes with the Literature

membrane	water permeability, A (L/(m2 h bar))	metal ion	rejection (%)	refs
dow membrane NF90	7.14	Pb2+ [Pb(NO3)2]	91–94	(58)
dow membrane NF270	13.2	Pb2+ [Pb(NO3)2]	∼60	(59)
		Cd2+ [Cd(NO3)2]	∼68
polybenzimidazole/polyethersulfone dual-layer hollow fiber membrane	0.826	Pb2+ [Pb(NO3)2]	93	(60)
		Cd2+ [Cd(NO3)2]	95
matrimid/PEI/Nexar	2.4	Pb2+ [Pb(NO3)2]	99.8	(61)
		Cd2+ [CdCl2]	98.2
PAN/SPEB blend	7.62	Pb2+ [Pb(NO3)2]	94.6	(62)
		Cd2+ [CdCl2]	95.1
PIP/PEI-Ag/H2N-NH2	8.0	Pb2+ [PbCl2]	99.6	(27)
tetrathioterephthalate/PES	10.0	Pb2+ [Pb(NO3)2]	97.4	(63)
ED-g-MWCNT/PES	8.0	Pb2+ [PbCl2]	90.5	(64)
PEI-CNCs	5.98	Pb2+ [PbCl2]	90.8	(65)
HTFC-1	7.0	Pb2+ [Pb(NO3)2]	98.1	this study
		Cd2+ [Cd(NO3)2]	96.3

(A) Pb2+ and pan class="Gene">Cd2+ rejection of the HTFC-1 membrane at pH 5, 150 psi (10.3 bar), and 20 °C. The filtration was performed with 10 ppm of Pb(NO3)2 and Cd(NO3)2 aqueous solution individually. (B) Antifouling performance of the control, HTFC-IPA, HTFC-1, HTFC-2, and HTFC-3 membranes with 200 ppm of aqueous humic acid (HA) as feed at 150 psi (10.3 bar) and 20 °C. Water permeability was measured in the first 8 h using DI water as a feed solution. Then, the feed solution was replaced with 200 ppm aqueous HA, and filtration was continued for another 8 h. Finally, the membranes were washed with DI water, and again the water permeability was measured for another 8 h using DI water as feed.

The rejection of higher metal ions (Pb2+ and Cd2+) of all of the as-prepared membranes was attributed mainly to the NF membrane surface charge. As all of the prepared NF membranes were positively charged at pH 5, the metal ions (Pb2+ and Cd2+) were rejected via electrostatic repulsion (Donnan effect). The positive charge on the control and HTFC-IPA NF membrane at pH 5 was attributed to the higher cross-linking density, which led to the reduced unavailability of the free carboxylic acid group. For HTFC-1, HTFC-2, and HTFC-3 membranes, the presence of EDA increased the positive charge. However, the membrane HTFC-3 membrane demonstrated slightly reduced metal ion rejection owing to the increased rp value (0.42 nm) compared to other NF membranes prepared.

Fouling of the NF membrane is one of the bottlenecks during membrane filtration. The presence of NOM, such as HA in the wastewater, will affect the NF membrane flux by adsorbing on the membrane surface.[66] In this study, the antifouling ability of all of the prepared NF membranes was analyzed for 24 h and is represented in Figure B. As shown, all of the NF membranes exhibited a sudden drop in water permeability when the feed solution was changed from water to HA solution. The sudden drop in water permeability was ascribed to the deposition of HA molecules on the membrane surface, which acted as a barrier for the water molecule in passing through the membrane. Then, the control and all HTFC NF membranes’ antifouling efficiencies were evaluated by measuring the flux recovery ratio (FRR) and are depicted in Figure S5. In general, the TFC NF membranes prepared with PIP as the monomer exhibited FRR in the range of 60–70%.[67] However, in this study, simple water washing of all of the NF membranes prepared using HP as a monomer bestowed FRR of >94%. When compared to all of the NF membranes prepared in this study, the HTFC-1 membrane demonstrated higher FRR (96.9%) and water permeability.

The improved antifouling performance of the as-prepared NF membranes can be explained based on the NF membrane surface physicochemical properties such as improved hydrophilicity and reduced surface roughness. All of the HTFC membranes conferred a contact angle of less than 24.1°, especially the HTFC-1 membrane, which exhibited a contact angle of 17.9° (Figure B). As these HTFC NF membranes are more hydrophilic in nature, they easily form a strong hydration layer on the membrane surface. The as-formed hydration avoided adsorption of foulant molecules (HA) on the membrane surface. Furthermore, the as-formed minimum amount of foulant molecules on the hydrophilic membrane can easily be removed by simple water washing. As shown in Figure B, the reduced surface roughness of the HTFC NF membranes (especially HTFC-1 Ra = 9.43 nm) is attributed to the reduced adsorption of foulant molecules (HA). Indeed, colloidal fouling in NF membranes can be directly related to surface roughness since these colloidal foulant molecules clog the valleys of the rough membrane surfaces.[68] Therefore, the reduced surface roughness of the as-prepared HTFC NF membranes could demonstrate improved antifouling performance.

Conclusions

In summary, we developed a new thin-film composite nanofiltration (NF) membrane using homopiperazine as a monomer. The influence of post-treatment with (EDA)/IPA on the NF membrane performances was studied. The as-prepared HTFC NF membranes exhibited low surface roughness, confirmed by AFM. The ζ-potential analysis revealed that the negatively charged PA layers of HTFC membranes became positively charged by simple EDA post-treatment. The optimized NF membrane prepared with 2 wt % HP, 0.15 wt % TMC, and post-treatment with EDA in IPA for 1 min demonstrated a pure water permeability of 7.0 ± 0.3 L/(m2 h bar) and salt rejections of 97.0 ± 0.6, 97.4 ± 0.5, 23.3 ± 0.6, 98.1 ± 0.3, and 96.3 ± 0.4% for Na2SO4, MgSO4, NaCl, Pb2+, and Cd2+, respectively. When fouled with HA, the NF membrane exhibited a flux recovery ratio of 96.9 ± 0.4% upon cross-flow washing. Overall, the as-prepared NF membranes are promising separation materials for brackish water desalination and the removal of heavy metal ions.

Experimental Section

Materials

Polysulfone (pan class="Chemical">PSF, Mn ∼22 000), homopiperazine (HP, 98%), sodium hydroxide (NaOH, >98%, pellets), lead nitrate (Pb(NO3)2, 99.9%), cadmium nitrate tetrahydrate (Cd(NO3)2·4H2O, 98%), humic acid sodium salt (HA), and ethylenediamine (EDA, >99%) were purchased from Sigma-Aldrich. 1,3,5-Benzenetricarboxylic acid chloride (TMC, 98%) was purchased from Acros. Poly(ethylene glycol) (PEG) of different molecular weights and N,N′-dimethylformamide (DMF, anhydrous, 99.8%) were purchased from Loba Chemie. Isopar-G was purchased from Univar. Poly(ethylene terephthalate) nonwoven fabric (PET, K#01 3249) was purchased from Hollytex. Isopropyl alcohol (IPA, >99.8%), sodium chloride (NaCl, >99.5%), anhydrous sodium sulfate (Na2SO4, >99.0%), and anhydrous magnesium sulfate (MgSO4, >98.0%) were obtained from Fisher Scientific.

Fabrication of Nanofiltration Membranes

The PSF beads were dried in a vacuum (∼25 in Hg) at 60 °C for 12 h to remove adsorbed pan class="Chemical">water. The PSF supports were fabricated through the nonsolvent induced phase separation (NIPS) method.[14] Briefly, 15 wt % PSF was dissolved in DMF and stirred for 8 h at 60 °C. The solution was deaerated by keeping the dope solution at room temperature for 6 h without stirring. The nonwoven PET fabric was secured on the glass plate by taping it from the backside and prewetted with DMF; the excess DMF was removed by Kimwipes (Kimberley-Clark). The dope solution was poured on the fabric and cast on the wet PET using a casting knife (Gardco) with an adjustable gap set at 50 mils. After the solution was cast, the substrates were immediately immersed in a coagulation bath containing water. After gelation (5 min), the membranes were transferred to another water bath and soaked for 24 h to remove the residual solvent. The as-prepared PSF support membrane exhibited a thickness of ∼100 μm measured by a micrometer at different locations. The PSF substrate membranes were stored in deionized water at 4 °C until use.

The poly(homopiperazine–amide) thin-film composite (HTFC) nanofiltration (NF) membranes were prepared by performing interfacial polymerization (IP) on the prepared PSF support. All of the TFC membranes were prepared at room temperature (20 °C) and relative humidity of 60%. In brief, PSF substrates were sandwiched between the glass plates and HDPE frames, creating wells 1 cm deep. Next, 30 mL of DI water containing 2 wt % HP and 0.35 wt % NaOH was poured on the substrate, and the substrate was allowed to rest for 2 min. The excess HP solution was drained by keeping the frame in the vertical position for 1 min. The residual droplets of the HP solution were removed by gently padding the support, using a Kimwipe. In the second step, 30 mL of Isopar-G solution containing 0.15 wt % TMC was poured into the well and left for 1 min to allow the reaction to complete. Subsequently, the excess solution was drained by keeping the frame in a vertical position for 1 min. The as-formed PA films were post-treated with 20 mL of IPA containing 1 wt % EDA. The duration of the post-treatment step was chosen as a variable. Membranes post-treated for 1, 2, and 3 min were labeled as HTFC-1, HTFC-2, and HTFC-3, respectively. Subsequently, the excess EDA solution was removed, and the membranes were cured in a convective oven for 8 min at 60 °C. We labeled the membranes that did not go through any postprocessing and one that was washed for 1 min with IPA as the control and HTFC-IPA, respectively. The prepared membranes were stored in deionized water at 4 °C until use. The synthetic scheme of the active layer of the NF membrane is presented in Scheme . The details of reactant concentrations and post-treatment are given in Table .

Table 3

Poly(Homopiperazine–Amide) Thin-Film Composite (HTFC) Membrane Parameters

membrane	HP (wt %)	TMC (wt %)	NaOH (g)	post-treatment
control	2	0.15	0.1	no post-treatment
HTFC-IPA	2	0.15	0.1	IPA washed
HTFC-1	2	0.15	0.1	1 wt % EDA in IPA for 1 min
HTFC-2	2	0.15	0.1	1 wt % EDA in IPA for 2 min
HTFC-3	2	0.15	0.1	1 wt % EDA in IPA for 3 min

HP, homopiperazine; TMC, 1,3,5-benzenetricarboxylic acid chloride; EDA, ethylenediamine; IPA, isopropyl alcohol.

HP, pan class="Chemical">homopiperazine; TMC, 1,3,5-benzenetricarboxylic acid chloride; EDA, ethylenediamine; IPA, isopropyl alcohol.

Characterization of HTFC NF Membranes

The transport properties of NF membranes were characterized using a custom-made cross-flow stainless steel membrane cell with an active area of 19 cm2. All of the measurement was done at 20 °C. The feed solution was circulated using a Hydra-Cell pump (Wanner Engineering Inc., Minneapolis, MN). The temperature of the feed solution was maintained using a VWR recirculating chiller. The cross-flow rate was monitored by a rotameter, and the volumetric flow rate (F) for the permeate was measured using a digital flow meter (Tovatech FlowCal 5000), connected to a PC. All of the experiments were replicated six times. For each measurement, before collecting water flux (Jw) and solute rejection, each membrane was compacted at 170 psi (11.7 bar) for 1 h using DI water, allowing for permeate flux to reach the steady-state condition. For each condition, Jw (L/(m2 h)) and water permeability (A) (L/(m2 h bar)) were measured at 150 psi (10.3 bar) using the following equations:where F is the permeate flow rate (L/h), Am is the effective surface area of the membrane in the module (m2), and Δp is the operating pressure (bar).

The solute rejection efficiency of the NF membrane for pan class="Chemical">NaCl, Na2SO4, and MgSO4 was measured using a feed solution with the dissolved solid concentration of 2000 ppm. All of the measurements were conducted at 150 psi (10.3 bar) and 20 °C. The ionic conductivities of both the feed and permeate were measured using a conductivity meter (Oakton CON 2700), calibrated using 0.01 M KCl standard solution. For the heavy metal removal efficiency calculations, Pb(NO3)2 and Cd(NO3)2 were used as salts. To evaluate the removal efficiency, 10 ppm aqueous solutions of these salts were used and the experiments were performed at 150 psi (10.3 bar), 20 °C, and pH 5. The metal ion concentrations, in both the feed and permeate, were determined using an inductively coupled plasma mass spectrometer (ICP-MS) (ThermoFisher iCAP RQ). The % rejection (R) was calculated using eq where Cp and Cf are the permeate and feed conductivities, respectively.

The molecular weight cutoff (MWCO) of the membrane wpan class="Chemical">as determined by filtering PEG molecules with different molecular weights (200, 400, 600, and 1000 Da) at 150 psi (10.3 bar) and 20 °C. The concentration of PEG in both the feed and permeate was estimated by measuring the total organic carbon (TOC) using the SHIMADZU TOC-L analyzer. The rejection (R) was calculated using eq , where Cp and Cf represent the concentrations of PEG in permeate and feed, respectively.

Accordingly, the PEG rejection curve wpan class="Chemical">as plotted, and the MWCO of the membrane was defined to be the equivalent PEG molecular weight at the rejection value of 90%.[69] The Stokes radius (rp) (nm) of the PEG was also determined according to eq (70)where MW is the molecular weight of the PEG used.

The antifouling performance of all of the HTFC membpan class="Chemical">ranes was evaluated using a feed solution containing 200 ppm of HA.[71] For this purpose, the membrane sample was loaded into the measurement cell and the water permeability (Jw0) was measured at 150 psi (10.3 bar) for 8 hours. Then, we replaced the feed solution with an aqueous solution containing 200 ppm HA and continued the filtration at the same condition for another 8 h. The permeate flux (Jp) was measured continuously. Once the membrane fouled with the HA solution, the feed solution was replaced with RO water. The fouled membrane was washed under the cross-flow condition, at 10 psi and 20 °C, to remove the loosely adhered HA molecules from the membrane surface. After the washing cycle, the water permeability (Jw) was measured at the same condition (10.3 bar, 20 °C) and the flux recovery ratio (FRR) was calculated using the following equation:

The detailed instrumental chapan class="Chemical">racterization of the HTFC NF membrane is given in the Supporting Information S1.