Limiting thickness of polyamide–polysulfone thin-film-composite nanofiltration membrane

Desalination 346 (2014) 19–29

Contents lists available at ScienceDirect

Desalination journal homepage: www.elsevier.com/locate/desal

Limiting thickness of polyamide–polysulfone thin-film-composite nanofiltration membrane Polisetti Veerababu a, Bhavik B. Vyas a, Puyam S. Singh a,b,⁎, Paramita Ray a,b a b

CSIR-Central Salt & Marine Chemicals Research Institute, RO Membrane Discipline, G. B. Marg, Bhavnagar, 364 002 Gujarat, India Academy of Scientific and Innovative Research (AcSIR-CSMCRI), G. B. Marg, Bhavnagar, 364002, Gujarat, India

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• A systematic study to explore limiting thickness of NF membrane. • Membrane thickness and surface roughness increased with support of decreased pore size. • High-flux membrane of 94 l·m− 2·h− 1 with moderate 84% MgSO4 rejection. • High-selective membrane of 92% MgSO4 rejection with moderate flux of 24 l·m− 2·h− 1.

a r t i c l e

i n f o

Article history: Received 22 November 2013 Received in revised form 5 May 2014 Accepted 6 May 2014 Available online 23 May 2014 Keywords: Thin-film-composite Nanofiltration Polyamide Thickness variation Performance

a b s t r a c t A systematic study was carried out to explore limiting thickness of the state-of-the-art nanofiltration membrane. The interfacial polymerization between aqueous solution of piperazine and organic solution of trimesoyl chloride over polysulfone ultrafiltration supports of average pore size 20, 100 and 200 nm, respectively had been performed to vary the membrane thickness and properties. The prepared membranes were designated as TFC-12, TFC-15 and TFC-24, respectively. Influence of the supports on membrane thickness, surface roughness and potential was observed by SEM, AFM, ATR-IR and zeta-potential measurements. The membrane thickness and surface roughness were found in increasing order of TFC-12 b TFC-15 b TFC-24. When tested for desalination of brackish water of 2000 ppm NaCl or 1000 ppm MgSO4 at 150 psig, TFC-24 exhibited the highest salt rejection efficiency but the least flux of 26 l·m−2·h−1 while the flux was enhanced to about 94 l·m−2·h−1 but with lesser salt rejection for TFC-12 which was in agreement with the differences in membrane thickness and roughness. The results indicated a trade-off performance relationship for the nanofiltration membranes that the membranes of high-flux with moderate selectivity or high-selectivity with moderate flux could be prepared by varying supports from a same preparation condition. © 2014 Elsevier B.V. All rights reserved.

1. Introduction The pressure-driven separation processes based on polyamide– polysulfone thin-film-composite (TFC) membranes are already successful approaches for satisfying water demands required for domestic, agricultural and industrial uses throughout the world. It is economical, ⁎ Corresponding author at: CSIR-Central Salt & Marine Chemicals Research Institute, RO Membrane Discipline, G. B. Marg, Bhavnagar, 364 002 Gujarat, India. Tel.: + 91 278 2567760; fax: +91 278 2567562. E-mail address: [email protected] (P.S. Singh).

http://dx.doi.org/10.1016/j.desal.2014.05.007 0011-9164/© 2014 Elsevier B.V. All rights reserved.

safe and an alternative to energy-intensive conventional processes. The present challenge is the preparation of the thin-film-composite membranes with a very thin active layer. This is of utmost importance as the fabrication of the membranes in the form of thin film as thin as possible over a porous support is highly desirable because the thin membrane film will allow optimal transport of molecules while the support will provide mechanical strength. A.P. Rao et al. [1] prepared several poly(m-phenylenediamine trimesamide)-polysulfone TFC membranes using semi-automatic casting and coating units in which the polyamide layer thickness could be varied from about 160 to 280 nm. It was observed on their study that dependence of water flux on polyamide film

20

P. Veerababu et al. / Desalination 346 (2014) 19–29

thickness as an inverse correlation trend, that, for the same salt rejection membranes, the permeate flux was more for those membranes which had a lesser film thickness. It was also observed from a detailed characterization studies performed on membranes of a large-scale preparation (1 m breadth × 90 m length) [2], that the polyamide film structure might have a range of network structures with varied macromolecular structural units depending on the ratio of linear polymer chain network having \COOH groups and cross-linked networks of \CONH-linkages, and that this structural variance in macromolecular chains mainly resulted in salt rejection variance which was in consistent with zeta-potential of membrane surface which showed a higher negative potential for the higher performing membrane that had polyamide network structure with more of pendant COOH group. The changes in the structure and composition of the polyamide film influenced the membrane performance both in terms of the salt rejection and water flux. While poly(m-phenylenediamine trimesamide)-

polysulfone TFC membranes are useful for RO process applications, the poly(piperazinetrimesamide)-polysulfone TFC membranes are useful for nanofilitration (NF) process applications. We reported earlier [3] that the different surface pore sizes (70 and 150 nm surface pore size average) of polysulfone support could affect the formation of the different types of TFC RO polyamide membranes. Similar studies of the porous polysulfone support effects on preparation of different TFC RO membrane types were also reported by others [4] in which the supports had average pore size in the range of 30–70 nm with different physicochemical properties. Further, it was reported [5] that the effect of supports of 40–90 nm pore size on preparation of different types of TFC NF membranes in which the top polyamide layer formed was about 1 μm thick. It exhibited flux of 15–25 l·m− 2·h− 1when tested for an aqueous feed of 1000 ppm Na2SO4 or MgSO4 with only a slight change in the salt rejection efficiency (~95% Na2SO4 and ~ 87% MgSO4 for all the membranes) even though there were differences in physical

C)

B)

A)

Fig. 1. Top surface SEM images of the TFC membranes, TFC-24 (A), TFC-15 (B) and TFC-12 (C).

P. Veerababu et al. / Desalination 346 (2014) 19–29

structures of the membranes. This could be due to the thick polyamide layer so that effects of supports on the salt rejection efficiency property of the membranes were negligible. Recently, we have published the fundamental and performance study of poly(piperazine trimesamide) based thin film composite NF membranes prepared in a large scale (0.5 m × 30 m scale) using inhouse membrane casting and coating units developed at CSIR-CSMCRI, Bhavnagar, India [6]. Polysulfone ultrafiltration membrane of about 70 nm surface pore size average was used as support in all the cases. Based on the different preparation conditions, the different NF membranes could be prepared, which exhibited NaCl rejection efficiency (%) ranging from 15 to 37%, MgSO4 rejection efficiency (%) ranging from 62 to 95% and the flux range of 32 to 73 l·m−2·h−1 when tested for an aqueous feed containing 1000 ppm NaCl/500 ppm MgSO4 at 150 psig. In the present paper, rather than preparation conditions, we use polysulfone ultrafiltration supports of different pore sizes to selectively vary the nature of the interfacial polymerization reactions leading to formation of NF polyamide films of different surface morphology and thickness and we follow the result using SEM, AFM, ATR-IR, zetapotential and performance measurements. In all of these TFC NF membranes, the preparation condition was kept the same and the membranes were prepared in a large-scale using the casting and coating units. The three polysulfone ultrafiltration supports chosen for the study had surface pore size average of (a) 20 nm, (b) 100 nm and (c) 200 nm, respectively. 2. Experimental 2.1. Preparation of the ultrafiltration polysulfone support membranes Polysulfone solution was prepared by dissolving a desired amount of dried (in vacuum oven at a temperature of 80 °C for 6 h) polysulfone (UDEL P-3500, Solvay Advanced Polymers, USA) in dimethyl formamide, DMF (Qualigen Fine Chemicals, India), in a round bottom flask at a temperature of 80 °C under constant stirring. The stirrer speed was maintained at about 1500 rpm. Three polymer solution concentrations of 12, 15 and 24% (w/w) were prepared. The total dissolution period

A)

was about 4–6 h. The polysulfone solution was kept in air tight condition at room temperature for 24 h to remove all the air bubbles. The polysulfone solution was then cast in film form on a non-woven polyester fabric (Nordyls TS 100 from Polymer group Inc. France) of width 0.3 m and length 30 m in continuous mode at a speed of 2 m/min with the help of in-house fabricated membrane casting machine. The thickness of the membrane was controlled by adjusting the gap between the casting blade and fabric with the help of a micrometer fixed at both ends. The casting unit was an enclosed chamber in which humidity was controlled at about 25% and temperature at 30 °C with the help of a humidifier. This nascent membrane which cast on the polyester fabric was then passed through a gelling bath at 25 °C made up of 100 L water, DMF (4%) and sodium lauryl sulfate (0.5%) where gelation of the polysulfone layer takes place by phase inversion process in a continuous mode. The resultant membrane was washed with running deionized water stream for a period of 20 min and preserved in deionized water for a period of 48 h for the removal of residual DMF. 2.2. Preparation, characterization and performance studies of the TFC NF membrane Thin film composite membrane was prepared in a continuous mode by using membrane coating machine as described in our earlier publication [6]. The ultrafiltration polysulfone membrane of length 30 m and width 0.3 m was mounted in the mounting roll. It was then passed through an aqueous bath containing piperazine (2 wt.%) and then organic (hexane) bath containing trimesoyl chloride (0.1 wt.%) at a desired speed controlled by the driving pulley. The dipping time in aqueous and organic bath was 45 s each. Interfacial polymerization reaction of piperazine and trimesoyl chloride on polysulfone membrane surface results in the formation of thin film composite poly (piperazinetrimesamide) membranes. Subsequently the membrane went through a curing chamber where it came in contact for 600 s with a hot plate maintained at 100 °C through dry air heating. The membrane was then thoroughly washed with warm water to remove unreacted amines, was treated with a mixture of glycerol and sodium lauryl sulfate, dried at a temperature of 50 °C, rolled using a receiving roller, and then

B) ~ 300 nm

21

C) ~ 200 nm

~ 100 nm

Fig. 2. Cross-section SEM images of the TFC membranes, TFC-24 (A), TFC-15 (B) and TFC-12 (C).

22

P. Veerababu et al. / Desalination 346 (2014) 19–29

preserved by covering with thick black plastic sheet. Two other membrane series were prepared at i) lower curing temperature of 90 °C and curing time of 300 s (A series); and ii) at lower PIP monomer concentration (1 wt.%), lower curing temperature of 90 °C and curing time of 180 s (B series). The surface chemistry of the membranes was studied by Attenuated Total Reflectance (ATR)-Infrared spectroscopy (IR) instrument which is a Barnes model 300 with continuously variable ATR accessory interfaced to a Nicolet 5DX Fourier transforms infrared spectrometer. For ATR-IR studies of our samples, germanium crystal at 45° angle of incidence was employed which gave probing depths of 0.39–0.67 μm in the chemical infrared region of interest. Dried membrane samples were cut according to ATR crystal size (6 cm × 2.2 cm) and mounted on both faces of the germanium crystal, the active layer facing the crystal surface. Spectra were recorded in the range 650–4000 cm−1. The membrane surface charge was estimated by measuring the streaming and zeta potential of the prepared NF membranes with the help of

Zeta CAD zeta potential analyzer supplied by CAD Instrumentation, France. For this, membrane samples were first equilibrated with electrolyte solution by soaking them for 12 h in the test solution. A streaming channel of defined dimension was then formed by mounting two identical flat membranes separated by a spacer with the active sides facing each other; each in one half of a rectangular cell. Electrolyte is forced through this slit channel using nitrogen gas, the pressure of which was controlled by a pressure controller. Electrolyte solution was circulated through the channel for a period of 45 min at a trans-membrane pressure for instrument stabilization and thus to start the experiment. The electrical potential developed due to this imposed movement of the electrolyte through this thin slit channel is sensed by two Ag/AgCl electrodes (placed one on each side i.e. at inlet and outlet of the channel). The electrodes were connected with a Keithley multi-meter to measure the electrical potential difference developed in the solution along the slit. The electrical potential difference (streaming potential) was measured alternatively in the two flow directions for continuously

A)

300 nm

B)

200 nm

C)

300nm

Fig. 3. Top surface SEM images of the polysulfone membrane supports, UF24 (A), UF15 (B) and UF12 (C). Inset: Magnified portion of the image.

P. Veerababu et al. / Desalination 346 (2014) 19–29

increasing pressure (from 0 to 500 mbar); the temperature, electric conductivity of the solution was also recorded by the instrument. The streaming potential coefficient i.e. the zeta potential was determined from the slope of the plot of potential difference versus pressure difference. The effect of pH on membrane surface potential was also studied in the range of 3–9. The pH of the aqueous solution was adjusted by adding required amount of HCl or NaOH. SEM images were recorded using LEO 1430VP scanning electron microscope at 15 kV accelerating voltage to examine surface morphology and thickness of the coated polyamide layer of the membranes. Membrane samples were gold coated by sputter coater for producing electrical conductivity Atomic force microscopy (AFM) images of the samples were acquired using NT-MDT AFM instrument. Pure water permeability of the membranes was studied using deionized water while the rejection profiles of the membranes were performed using aqueous feed of NaCl (1000 ppm) or MgSO4 (500 ppm) at 150 psi. The membrane sample test coupons were evaluated for their selectivity and flux performances on a batch type RO test kit comprising four cells in series in which the feed flow was in the cross-flow manner. The concentration of the inorganic salts in feed and permeate were estimated by a standardized conductivity meter as well as inductively coupled plasma spectroscopy, ICP Optima 2000 DV model.

23

skin layer was thinner, leading to the possibility of higher degree of defects, more porous and, consequently, lower salt rejection efficiency (65–70% NaCl) but higher flux of about 80 l·m− 2 ·h− 1 at 250 psig. Ghosh et al. [4] had also observed the different properties of laboratory-scale TFC membranes supported on hand-cast and commercial polysulfone supports with pore sizes in the range of 30–70 nm of different surface properties having water contact angles between 60° and 80°. It can be noted, however, the flux range 4– 25 l·m− 2·h− 1at 145 psig (1 Mpa) and NaCl salt rejection efficiency (75–91%) observed by them was relatively less as compared to the performance of the controlled, automated machine coated TFC RO

A)

3. Results & discussion 3.1. SEM study of the TFC NF membranes The three TFC NF poly (piperazine trimesamide) membrane types (TFC-NF24, TFC-NF15 and TFC-NF12) which were supported on different ultrafiltration polysulfone membranes were compared in terms of their polyamide surface microstructure and thickness. The polyamide surface microstructure for the TFC membrane types as observed by SEM is shown in Fig. 1. A typical globular-like nodular morphology of the polyamide was clearly visible from the SEM surface images in the membranes TFC-NF24 and TFC-NF15. The nodule size was about 100–200 nm for TFC-NF24 and about 50–100 nm for TFC-NF15. The nodule size of the TFC-NF12 membrane was not very distinctive from the membrane surface but it appeared to be smaller than the nodules of other types. The cross-section SEM images of the membranes are shown in Fig. 2. The TFC-NF24, TFC-NF15 and TFC-NF12 membranes had polyamide layer thickness of about 300, 200 and 100 nm, respectively over the polysulfone supports. The observed differences in polyamide surface microstructure morphology and thickness among the TFC types from the same preparation condition indicated that the polysulfone support of different pore size influenced on polyamide formation of piperazine-trimesoyl chloride interfacial polymerization. The impact of the support membrane on the formation of TFC membranes can be different depending upon the physico-chemical properties of the support and coating conditions. Furthermore, the membranes prepared from the large-scale preparation using automated controlled coating were different from the smallscale tray experiments. The small-scale results could lead to various errors in development of pilot and full-scale systems as noted by others [7,8] and from our own experience in large-scale TFC membrane development [9]. Therefore, the influence of polysulfone support on formation of different TFC membrane types is required to be discussed in line with the observations from the specific preparation systems and conditions. We had shown in our earlier publication [3] that the influence of two polysulfone ultrafiltration supports having pore size average of 70 and 150 nm on large-scale prepared TFC RO membranes that suggested a two-fold thicker polyamide layer formation due to reduced penetration of polyamide in the smaller-pore support exhibiting excellent salt rejection efficiency (rejection average of 96% NaCl). Whereas in the case of the other TFC membrane having larger-pore support the polyamide

B)

15 µm

C)

Fig. 4. Cross-section SEM images of the polysulfone membrane supports, UF24 (A), UF15 (B) and UF12 (C). Total thickness of UF24 was about 25 μm while the thicknesses of others were about 15 μm each. Images were acquired after separation of the polysulfone from bottom reinforcing polyester fabric.

24

P. Veerababu et al. / Desalination 346 (2014) 19–29

Table 1 Properties of membrane supports. Polysulfone

UF12

UF15

UF24

Thickness (μm) Cross-sectional morphology

15 Asymmetric pore morphology (smaller surface pores and large macro-voids underneath) 540 82 500 200 16 29

15 Gradient pore morphology (smaller surface pores with gradual increase underneath) 364 61 200 100 14 19

20 Gradient pore morphology (smaller surface pores with gradual increase underneath) 79 44 100 20 10 14

Pure water flux (l·m−2·h−1) at 50 psig Porosity (%)a MWCO, KDa Surface pore size averageb (nm) Flow-through average pore sizec (nm) Size-exclusiond (nm)

a Porosity (%) = (Wwet − Wdry) × 100/(A × d × ρ), where Wwet and Wdry are the weight of the membranes in wet and dry conditions respectively, A is the membrane area, d is the density of water and ρ is the thickness of the membrane. b From top surface SEM image. c Based on Hagen–Poiseuille equation. d Based on pore diameter (size) = 0.09 × MWCO0.44.

membranes [1–3,9]. Such low flux and selectivity of the TFC RO membrane was also observed by us in manually coated TFC RO membrane [10]. The performance variation of the TFC membrane types could be

due to the reason that the polyamide layer formed through interfacial polymerization from different preparation systems was not necessarily homogeneous, but could have areas of porosity as observed systematically in free-standing polyamide film of 50 μm thickness [11]. 3.2. Properties of the TFC NF membrane supports

C)

B)

A)

Fig. 3 shows SEM surface images of polysulfone supports bearing sample numbers UF24, UF15 and UF12. These support membranes were used for the preparation of TFC-24, TFC-15 and TFC-12 membranes, respectively. As shown in Fig. 3, the top surface SEM image of the UF24 had relatively denser surface morphology comprised of small pores (~20 nm average size) while the UF15 and UF12 supports exhibited the surface pore size average of about 100 nm and 200 nm, respectively. These polysulfone supports of different pore sizes were prepared from polysulfone solution by the phase inversion process using automatic in-house membrane casting unit as reported elsewhere [6]. The UF12, UF15 and UF24 supports of different pore sizes were achieved by casting from polymer solution of 12, 15 and 24% (w/w) in DMF respectively. In the phase inversion process, there is a diffusionexchange of solvent and non-solvent resulting in a thermodynamic instability of the polymer solution which in turn separates into solutions of polymer lean and polymer rich phases consequently to the formation of a denser layer of smaller pores from the polymer rich phase on top of a largely porous layer from the polymer lean phase. Thus, different membrane pores could be obtained by changing casting conditions (i.e., temperature of casting solution, composition and temperature of coagulating bath, type of polyester fabric, gate height of the casting blade etc.) while keeping the same polymer solution concentration due to the phenomenon as discussed above. The cross-sectional SEM images of these supports are shown in Fig. 4. A denser skin layer of smaller pores with gradual increase in pore sizes underneath was observed in all the support types. The densest skin layer in the case of UF24 and the most loose skin layer in the case of UF12 were observed which agreed well with the observed surface pore sizes as discussed above. Further, in the case of UF12 there was presence of large macrovoids beneath the skin layer. The comparison of these membrane supports is given in Table 1. The thickness of the UF12 and UF15 membranes was about 15 μm while the UF24 membrane was slightly thicker Table 2 Surface roughness properties of the TFC NF membranes.

Fig. 5. Top surface AFM images of the TFC membranes, TFC-24 (A), TFC-15 (B) and TFC-12 (C).

Properties

TFC-12

TFC-15

TFC-24

Peak to peak maximum distance (Sy), nm Mean value of peak to peak distance (μ), nm Average arithmetic roughness (Sa), nm Root mean square roughness (Sq), nm

77.6 44.7 7.4 9.4

179.6 70.8 17.9 22.3

708.1 368.7 105.1 131.1

P. Veerababu et al. / Desalination 346 (2014) 19–29

at about 20 μm. The pure water flux (540 l·m−2·h−1) was highest in case of the UF12 which was about 1.5 times of the flux for the UF15 and about 7 times of the flux for the UF24. Considering Hagen–Poiseuille equation [12], the calculated pore size from the pure water flux data for these membranes was found to be in the range 10–16 nm which was slightly smaller than the values obtained from the empirical relation of pore diameter and molecular weight cut-off (MWCO), pore diameter = 0.09 × MWCO0.44 [12]. These calculated pore size values were however much smaller than the observed 20–200 nm surfacepore size averages by top surface SEM images. This implied that the pore sizes observed on the surfaces were not necessarily flow-through pores, rather majority of these pores could be constricted or dead-end pores.

25

1200 nm for TFC-24. In accordance with the height scale of nodules, it was observed that surface roughness values (ΔZ) were smallest in the distribution range of 1–40 nm with median peak at about 10 nm in the case of TFC-12 membrane, while the roughness was in the range of 10–100 nm with median peak at 40 nm for TFC-15 membrane. TFC24 membrane surface exhibited wide distribution of surface roughness ranging from 20 to 300 nm. The results were again in accordance with the nodule sizes obtained from the image analysis. The histograms of the nodule sizes for the membranes are shown in Fig. 7. TFC-12, TFC15 and TFC-24 had size range 10–60 nm, 100–250 nm and 100– 450 nm, respectively. Both the TFC-12 and TFC-15 membranes had symmetric distribution of the nodule sizes with median size at about 40 and 180 nm, respectively; whereas TFC-24 exhibited wide distribution of the nodule sizes. (See Fig. 6.)

3.3. TFC NF membrane surface characterization by AFM study 3.4. ATR-IR study of the TFC NF membranes Surface AFM images of the TFC NF membranes are shown in Fig. 5. The surface morphology of the sample as revealed by the AFM image shows a typical nodular morphology. The nodule size average was biggest for TFC-24 while it was smallest for TFC-12. The surface roughness parameters in terms of the peak to peak maximum distance, the mean value of peak to peak distance, the average arithmetic roughness and the root mean square roughness were analyzed from the images and the results are given in Table 2. It can be seen from Table 2 that surface roughness is in an increasing order of TFC-12 b TFC-15 b TFC-24. The AFM images shown in Fig. 5 were further analyzed in terms of maximum and minimum heights (Z-scale) of the hill-ridge nodule morphology using Windows based NT-MDT Image Analysis Software Build 3.5.0.2064. The difference in height (ΔZ) could be considered as localized surface roughness of the nodules. The surface of TFC-12 membrane had the lowest height of all nodules. The maximum heights of the nodules for this membrane were in 20–60 nm range with median peak at about 50 nm. For TFC-15 the maximum height range and median peak were 40–120 nm and 90 nm while they were 800–1400 nm and

8

10

6

6 4

4

4 2 0

0

0 0

20

40

60

0

80

20

B)5

40

60

0

80

(i)

(ii)

3 2

0

0

0 0

20

40

60

80

100

0

120

20

Z-maximum (nm)

A)8

4 2

1

1

40

60

80

0

2

Counts

Counts

4

4

(ii)

6 4 2

0 400

800

Z-maximum (nm)

1200

120

(iii)

3 2 1 0

0 0

80

Δ Z (nm)

8

6

40

Z-minimum (nm)

(i)

80

(iii)

6

Counts

Counts

2

60

8

4

3

40

Δ Z(nm)

5

4

20

Z-maximum (nm)

Z-maximum (nm)

Counts

6

2

2

(iii)

8

(ii)

Counts

(i) Counts

Counts

C)8

Counts

Four samples from different portions of each membrane type were measured for sample uniformity but only one representative spectrum for each membrane type was considered in the comparison plot of ATR-IR spectra as shown in Fig. 8. The IR intensities of surface layers of TFC samples were definitely different from one another. The strongest absorbance of IR peaks was observed in the case of TFC-12 but least IR absorbance was for TFC-24. The IR absorbance intensities for the samples in the spectral frequency range of 1015–1730 cm−1 are given in Table 3. The relative IR intensities of the TFC samples are in an increasing order of TFC-24 b TFC-15 b TFC-12 suggesting that there are significant differences in the structures of the polyamide–polysulfone composite layers. The IR intensities along with the spectral assignments of polyamide and polysulfone are separately grouped in Table 3. The band at 1650–1670 cm − 1 is characteristic of amide I (C_O stretch) and the 1530–1540 cm−1 band is ascribable to C\N stretch of amide and were only clearly distinguishable in the TFC-24 but these peaks were either weak or indistinct for TFC-12 and TFC-15. Whereas, the 1730 cm−1

0

400

800

Z-minimum (nm)

1200

0

50

100 150 200 250 300

Δ Z (nm)

Fig. 6. Height (Z-scale) histograms of the AFM images (shown in Fig. 5) of the TFC membranes, TFC-24 (A), TFC-15 (B) and TFC-12 (C); Z-maximum (i), Z-minimum (ii) and Δ Z, Z-maximum–Z-minimum (iii).

26

P. Veerababu et al. / Desalination 346 (2014) 19–29

20

TFC-UF12 TFC-UF15 TFC-UF24

0.12

C)

Counts

15 10

0.10

Absorbance (a.u.)

5 0 20

40

60

80

Average size (nm) 16

0.08

0.06

Counts

12

B) 8

0.04

4

0.02

0 0

100

200

300

400

Average size (nm) 1000

1200

1600

1800

Fig. 8. ATR-IR spectra of the TFC membranes. The spectra are shifted in vertical direction by arbitrary units to present each of them distinctly.

4

Counts

1400

cm-1

5

A)

3 2 1 0 0

100

200

300

400

500

600

Average size (nm) Fig. 7. Histograms of the average nodule sizes of the TFC membranes analyzed from the surface AFM images, TFC-24 (A), TFC-15 (B) and TFC-12 (C).

band due to carboxylic acid (C_O) and the 1445 cm−1 band due to OH deformation band (acid) were distinctive for all the membrane types. The IR band (1730 cm−1) characteristic of carboxylic acid group could be considered to correlate with the amount of surface negative charge of the membrane based on the relationship between the IR peak intensity and COO− negative charge of \COOH groups. The IR (carboxylic acid) intensity of 1730 cm−1 band was relatively weak for TFC-12 as compared to the values for TFC-15 and TFC-24. The change in the intensity values was not compatible with the polyamide thickness values observed by SEM cross-section images. Such incompatibility between the polyamide thickness and IR intensity data had been observed in case of TFC RO membranes [3]. On the other hand, characteristic intensities of polysulfone were found to be the highest for TFC-12 among the membranes. The relative intensities between the polyamide and polysulfone support for the membranes were considered to compare the polyamide thickness coated over the support. The bands at 1247 and 1588 cm−1 which are the 2 strongest bands of the polysulfone and the polyamide characteristic band at 1730 cm−1 were chosen for the calculation. The relative intensity ratios of polyamide band (I1730) to polysulfone band (I1247 or I1588) are given in Table 3. The relative intensity ratio was the highest for TFC-24 which was about twice the ratio of TFC-15 while TFC-12 exhibited one-third the ratio of TFC-15. This means that the polyamide layer thickness of the membrane is increased in the order of TFC12 b TFC-15 b TFC-24. Further, the phenomenon of interfacial polycondensation to the formation of polyamide could be influenced by various

factors, such as types of reactant monomers, reaction rate, and impacts of different membrane supports, as described below. In 1976, Enkelmann and Wegner [13] expressed the limiting thickness of polyamide film growth based on the diffusion of amine monomers and water toward organic phase containing acid halide, and the competition between the amidation and hydrolysis reactions. Later, several other models on film growth such as, unlimited growth of film thickness with time [14,15], a finite film thickness at infinite time [16], and the limiting thickness and asymmetric distribution of density and charge [17] were reported. Of late, Freger [18] presented an approximate analytical model of film formation by interfacial polycondensation using fixed parameters of monomer diffusivities, reaction rate constants, monomer concentrations and thickness of the unstirred layer. Two or three different kinetic regimes in succession were assumed in the process in which each regime generates a different film structure pattern because of differences in their kinetic processes. The analysis of model suggests a direct correlation between the roughness and the

Table 3 IR absorbance values for the 3 types of polysulfone–polyamide composite membranes. Frequency (cm−1)

Spectra assignments

Polysulfone 1152 C\SO2\C symmetric stretching 1247 C\O\C stretching 1292 S_O stretching 1321 C\SO2\C asymmetric stretching 1487 CH3\C\CH3 stretching 1588 C_C aromatic ring stretching 1622 Aromatic ring breathing Polyamide 1445 OH deformation band (acid) 1530–1540 C\N stretching (amide) 1650–1670 C_O stretching (amide) 1730 C_O stretching (acid) IR absorbance intensity ratio I1730/I1247 IR absorbance intensity ratio I1730/I1588

IR (absorbance) TFC-24

TFC-15

TFC-12

0.0130 0.0170 0.0050 0.0040 0.0070 0.0120 0.0070

0.0196 0.0389 0.0106 0.0090 0.0211 0.0285 0.0212

0.0495 0.0858 0.0296 0.0219 0.0459 0.0520 0.0522

0.0030 0.0040 0.0060 0.0080 0.4705 0.6667

0.0090 Weak 0.0060 0.0108 0.2776 0.3789

0.0282 Indistinct Indistinct 0.0054 0.0629 0.1038

P. Veerababu et al. / Desalination 346 (2014) 19–29

thickness of the film. Using synchrotron SAXS [19], we suggested a tier structure comprised of molecular packing, primary compacted chains and clusters of primary units for RO and NF polyamide membrane. The nature of polymer chain compaction and cluster size may be caused by the differences in the reaction rates and the cross-link density of the polymer. Ramon et al. [20] had shown theoretically that the permeability of the RO/NF polyamide membranes strongly depends on the skin layer pore structure of the support membrane on which the coating film was formed. The larger pores of the polysulfone support used in TFC-12 membrane preparation could favor effective penetration of the piperazine monomer into the pores which subsequently reacted with the acid chloride inside the porous support matrix while the smaller pores of the support of TFC-24 may make it relatively difficult for polyamide formation inside the pores. Thus, for TFC-24, it resulted mainly in formation of a polyamide skin layer over the polysulfone and signal intensities are overall weaker for polysulfone due to thick layer of polyamide layer within the penetration depth of the IR beam. The relative intensity ratios of polyamide characteristic band (I1730) to polysulfone characteristic band (I1247 or I1588) (Table 3) decreases for the membrane coated on support of bigger pore size which indicates that more polyamide is coated on the support of smaller pore size. A schematic diagram on the illustration of TFC NF membranes of different thickness over the supports of different pore sizes are shown in Fig. 9.

3.5. Membrane performance and zeta-potential measurements The performances of the membranes were measured at 150 psi in terms of rejection efficiency of NaCl and MgSO4 and permeate flux. Eight samples were analyzed for each membrane type to generate

27

Table 4 Performance comparison of the membranes in desalination of two different feeds of saline water; one feed containing 2000 ppm NaCl and the other feed containing 1000 ppm MgSO4. Membrane performance Feed: 1000 ppm MgSO4 in water Average flux (l · m−2 · h−1) Standard deviation (flux) Salt Rejection (%), S/R Standard Deviation (S/R) Feed: 2000 ppm NaCl in water Average flux (l · m−2 · h−1) Standard deviation (flux) Salt Rejection (%), S/R Standard Deviation (S/R) Rejection ratio, MgSO4/NaCl

TFC-24

TFC-15

TFC-12

25.5 7.9 92.3 2.9

58.2 9.0 87.0 2.6

94.4 5.7 84.6 3.1

21.5 6.2 47.8 4.1 1.9

59.7 12.3 35.5 2.3 2.5

85.0 6.6 44.6 6.3 1.9

Conditions: Operating pressure = 150 psi, temperature = 25 °C.

statistical data of average values along with standard deviation. These average flux-rejection results along with the standard deviation are given in Table 4, whereas the overall performance data of all the samples are plotted as shown in Fig. 10. From the performance data, it was clear that there was significant difference among the TFC NF membranes formed on the supports of different pore sizes. With the change in the support pore size (surface) average from 20 to 200 nm the membrane flux was enhanced to about 94 l·m−2·h−1with MgSO4 rejection (85%) and NaCl rejection (45%) at 150 psig (as observed for TFC-12 membrane coated on UF-12 of 200 nm surface pore size average). Conversely, TFC-24 membrane coated on UF-24 of 20 nm surface pore size average had exhibited the highest salt rejection efficiency (92% MgSO4, 50% NaCl) but the least flux of 26 l·m−2·h−1 at 150 psig among the membranes.

Support

Pore size

Polyamide layer

Organic phase (Trimesoyl chloride)

Aqueous phase (Piperazine)

Pore size

Pore size

Fig. 9. Illustration of TFC membranes of different thickness over the supports of different pore sizes.

28

P. Veerababu et al. / Desalination 346 (2014) 19–29

100

B) 10

80 0 60

40

TFC-UF24 TFC-UF15 TFC-UF12

mV

-10

Flux (LMH) S/R (%) -

20

-20

0

TFC-24

TFC-15

TFC-12 -30

100

A) -40

Flux (LMH) S/R (%) -

80

3

4

5

6

7

8

9

10

pH 60

Fig. 11. Zeta-potentials values of the membranes. 40

20

0

TFC-24

TFC-15

TFC-12

Fig. 10. Performance comparison of the three membrane types in desalination of feed containing 2000 ppm NaCl (A) and the other feed containing 1000 ppm MgSO4 (B).

By using the three supports, the membranes were also prepared by varying monomer concentration, cure temperature and time. The results are given in Table 5. TFC-24A is characterized by the lowest flux but the highest salt rejection among the three samples, TFC-24A, TFC15A and TFC-12A. Similarly, TFC-24B exhibited the lowest flux but the highest salt rejection among the TFC-24B, TFC-15B and TFC-12B samples. Both these series of membranes were prepared at i) lower curing temperature of 90 °C and curing time of 300 s; and ii) at lower PIP monomer concentration (1 wt.%), lower curing temperature of 90 °C and curing time of 180 s as compared to TFC-24, TFC-15 and TFC-12 samples. Thus, it is evident from the results (Table 5) that the porous support influences the performance of the final membranes. The difference in surface properties among the types of NF membranes was also observed by zeta-potential values of the membranes surfaces. As shown in Fig. 11, the negative values of zeta-potential were in an increasing order of TFC-12 b TFC-15 b TFC-24. The differences in the trend of the zeta values could be due to change in charge of surface chemical functional groups on the influence of electrolyte.

The maximum negative zeta values of TFC-24 indicated that this surface could have relatively more amount of pendant COOH groups. The variation of the salt rejection efficiency (%RMgSO4 and %RNaCl) for the membranes may be primarily due to differences in the membrane surface potential as observed by the zeta potential measurements. However, concentration polarization may also influence the salt rejection efficiency [21]. On the other hand, the variation of the flux is found to be consistent with the membrane thickness. 4. Conclusion Thin-film-composite nanofiltration membranes were prepared by coating polyamide on polysulfone ultrafiltration supports (length 30 m and width 0.3 m) of average pore size 20, 100 and 200 nm, respectively in a continuous mode by using a motorized coating machine. For each polyamide coating, the condition of interfacial polymerization between aqueous solution containing piperazine (2 wt.%) and organic (hexane) solution containing trimesoyl chloride (0.1 wt.%) was same in which the reaction time was 45 s at 25 °C and subsequently cured at 100 °C for 10 min. Influence of the supports on membrane thickness, surface roughness and potential was observed by SEM, AFM, ATR-IR and zeta-potential measurements. It is suggested that the larger pores of the polysulfone support could favor effective penetration of the piperazine monomer into the pores which subsequently reacted with the acid chloride inside the porous support matrix while the smaller pores of support may make it relatively difficult for polyamide formation inside the pores. This could result into formation of a thin and thick polyamide

Table 5 Performance comparison of series of membranes prepared at different conditions in desalination of saline water at 150 psi and 25 °C. Membrane

TFC-24 TFC-15 TFC-12 TFC-24A TFC-15A TFC-12A TFC-24B TFC-15B TFC-12B

PIP, wt.%

TMC, wt.%

Reaction time (s)

Cure temp. °C

Cure time (s)

Flux

0.2 0.2 0.2 2.0 2.0 2.0 1.0 1.0 1.0

0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1

45 45 45 45 45 45 45 45 45

100 100 100 90 90 90 90 90 90

600 600 600 300 300 300 180 180 180

21.5 59.7 85.0 25.3 64.2 90.2 26.3 70.2 92.2

% salt rejection

NaCl

NaCl

MgSO4 ± ± ± ± ± ± ± ± ±

6.2 12.3 6.6 4.5 2.4 3.4 5.4 10.5 5.2

25.5 58.2 94.4 26.5 63.5 92.2 27.4 65.3 95.2

± ± ± ± ± ± ± ± ±

7.9 9.0 5.7 4.5 3.6 5.9 5.6 11.3 2.4

47.8 35.3 44.6 35.3 27.4 26.7 37.2 27.5 27.3

MgSO4 ± ± ± ± ± ± ± ± ±

4.1 2.3 6.3 5.2 5.2 4.5 2.6 6.7 4.6

92.3 87.0 84.6 80.2 79.4 77.6 79.1 78.3 76.2

± ± ± ± ± ± ± ± ±

2.9 2.6 3.1 6.2 6.3 5.2 3.4 5.2 3.7

P. Veerababu et al. / Desalination 346 (2014) 19–29

skin layer over the polysulfone supports of smaller and larger pores, respectively. When tested for desalination of brackish water of 2000 ppm NaCl or 1000 ppm MgSO4 at 150 psi, the thickest membrane exhibited the highest salt rejection efficiency but the least flux of 26 l·m−2·h−1 while the flux was enhanced to about 94 l·m−2·h−1 but with lesser salt rejection for the thinnest membrane which was in agreement with the differences in membrane thickness and roughness. The study reflects new dimension to understand the preparation of nanofiltration membranes with optimal membrane performance in terms of separation-selectivity and productivity-flux by controlling the top active-layer thickness and surface roughness. Nanofiltration membranes of high-flux with moderate selectivity or high-selectivity with moderate flux could be prepared by varying supports from a same preparation condition. Acknowledgments Financial assistance as research grants from the Council of Scientific & Industrial Research (9/1/CS/CSMCRI(1)/2012-13-PPD) and Ministry of Water Resources (29/INCGW-05/2010-R&D/2997-2006), Government of India as well as the instrumentation facility provided by Analytical Discipline & Centralized Instrument Facility, CSIR-CSMCRI, Bhavnagar, are gratefully acknowledged. References [1] A.P. Rao, S.V. Joshi, J.J. Trivedi, C.V. Devmurari, V.J. Shah, Structure-performance correlation of polyamide thin film composite membranes: effect of coating conditions on film formation, J. Membr. Sci. 211 (2003) 13. [2] P.S. Singh, A.P. Rao, P. Ray, A. Bhattacharya, K. Singh, N.K. Saha, A.V.R. Reddy, Techniques for characterization of polyamide thin film composite membranes, Desalination 282 (2011) 78. [3] P.S. Singh, S.V. Joshi, J.J. Trivedi, C.V. Devmurari, A.P. Rao, P.K. Ghosh, Probing the structural variations of thin film composite RO membranes obtained by coating polyamide over polysulfone membranes of different pore dimensions, J. Membr. Sci. 278 (2006) 19.

29

[4] A.K. Ghosh, E.M.V. Hoek, Impacts of support membrane structure and chemistry on polyamide–polysulfone interfacial composite membranes, J. Membr. Sci. 336 (2009) 140. [5] N. Misdan, W.J. Lau, A.F. Ismail, T. Matsuura, Formation of thin film composite nanofiltration membrane: effect of polysulfone substrate characteristics, Desalination 329 (2013) 9. [6] P.S. Singh, P. Ray, P. Kallem, S. Maurya, G.S. Trivedi, Structure and performance of nanofiltration membrane prepared in a large-scale at CSIR-CSMCRI using indigenous coating unit, Desalination 288 (2012) 8. [7] H.R. Rabie, P. Côté, N. Adams, A method for assessing membrane fouling in pilot and full-scale systems, Desalination 141 (2001) 237. [8] T. Schipolowski, A. Jezowska, G. Wozny, Reliability of membrane test cell measurements, Desalination 189 (2006) 71. [9] R. Rangarajan, N.V. Desai, S.L. Daga, S.V. Joshi, A.P. Rao, V.J. Shah, J.J. Trivedi, C.V. Devmurari, K. Singh, P.S. Bapat, H.L. Raval, S.K. Jewrajka, N.K. Saha, A. Bhattacharya, P.S. Singh, P. Ray, G.S. Trivedi, N. Pathak, A.V.R. Reddy, Thin film composite reverse osmosis membrane development and scale up at CSMCRI, Bhavnagar, Desalination 282 (2011) 68. [10] G.L. Jadav, P.S. Singh, Synthesis of novel silica-polyamide nanocomposite membrane with enhanced properties, J. Membr. Sci. 328 (2009) 257. [11] J. Lee, A. Hill, S. Kentish, Formation of a thick aromatic polyamide membrane by interfacial polymerisation, Sep. Purif. Technol. 104 (2013) 276. [12] M. Mulder, Basic Principles of Membrane Technology, 2nd edition Kluwer Academic Publishers, 1996. [13] V. Enkelman, G. Wegner, Mechanism of interfacial polycondensation and the direct synthesis of stable polyamide membranes, Makromol. Chem. 177 (1976) 3177. [14] L.J.J.M. Janssen, K. te Nijenhuis, Encapsulation by interfacial polycondensation-I. The capsule production and a model for wall growth, J. Membr. Sci. 65 (1992) 59. [15] L.J.J.M. Janssen, K. te Nijenhuis, Encapsulation by interfacial polycondensation-II. The membrane wall structure and the rate of wall growth, J. Membr. Sci. 65 (1992) 69. [16] J. Ji, J.M. Dickson, R.F. Childs, B.E. McCarry, Mathematical model for the formation of thin film composite membranes by interfacial polymerization: porous and dense films, Macromolecules 33 (2000) 624. [17] V. Freger, S. Srebnik, Mathematical model of charge and density distributions in interfacial polymerization of thin films, J. Appl. Polym. Sci. 88 (2003) 1162. [18] V. Freger, Kinetics of film formation by interfacial polycondensation, Langmuir 21 (2005) 1884. [19] P.S. Singh, P. Ray, Z. Xie, M. Hoang, Synchrotron SAXS to probe cross-linked network of polyamide ‘reverse osmosis’ and ‘nanofiltration’ membranes, J. Membr. Sci. 421–422 (2012) 51. [20] G.Z. Ramon, M.C.Y. Wong, E.M.V. Hoek, Transport through composite membrane, part 1: is there an optimal support membrane? J. Membr. Sci. 415–416 (2012) 298–305. [21] S. Bason, V. Freger, Phenomenological analysis of transport of mono- and divalent ions in nanofiltration, J. Membr. Sci. 360 (2010) 389–396.