Modification of polyamide membranes by

Journal of Membrane Science 546 (2018) 165–172

Contents lists available at ScienceDirect

Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci

Modification of polyamide membranes by hydrophobic molecular plugs for improved boron rejection Shiran Shultz, Maria Bass, Raphael Semiat, Viatcheslav Freger

T

⁎

Technion – Israel Institute of Technology, Wolfson Department of Chemical Engineering, Technion City, 32000 Haifa, Israel

A B S T R A C T Commercial polyamide RO membranes, though effective in terms of flux and salt removal, have a few drawbacks, in particular, poor rejection of boron (B), thus they are often unable to remove B to the required level in a one-pass sea desalination process. This complicates the process and available solutions increase the cost of desalinated water significantly. Here we explore in-situ modification procedure that can significantly increase their boron rejection by incorporating suitable modifying molecules in the selective polyamide layer. We propose that aliphatic amines be used as such “plug” molecules that combine a bulky hydrophobic moiety with a reactive group that can chemically or physically bind to polyamide layer, tighten its structure and increase selectivity. Based on previous results, we hypothesize that hydrophobicity of the selective layer increased by the immobilized amine “plug”, along with reduced pore size, may help disrupt water-boric acid association and decouple water and boron permeation. The results show that the proposed treatment with sufficiently long (up to 10–12 carbons) aliphatic amines, using either covalent bonding or simple sorption, may indeed reduce the boron passage by a factor of 2–4, without impairing the salt rejection of the membrane. The improved selectivity comes at the expense of some flux reduction, but the flux-selectivity tradeoff improves compared with commercial polyamide membranes.

1. Introduction Synthetic membranes for separation are used in many industrial applications. The main areas where membranes are in use are desalination of sea and brackish water and purification of potable water, industrial and municipal waste effluents. In reverse osmosis (RO) a semi-permeable membrane is employed to reject dissolved constituents (salts and pollutants) in the feeding water and allow water to pass through [1]. RO membrane desalination has become one of the main techniques, estimated as 44% from the total desalinated volume in the world, and is expected to grow in the coming years [2]. The progress in RO technology is greatly dependent on the development of RO membranes, as the membrane plays a key role in technological and economic efficiency of the RO process [3]. Thin-film composite (TFC) membranes with a polyamide top layer are the most common type of RO membranes used nowadays. The selective thin and dense aromatic polyamide top layer is prepared by interfacial polymerization of aromatic triacid and diamine and contains some unreacted, mainly, acidic carboxylic acid groups within the membrane and on the surface. This polyamide layer is typically 20–200 nm thick and provides the separation selectivity [4]. Compared

⁎

with older-type cellulosic membranes, the TFC aromatic polyamide membrane exhibits superior water flux and salt rejection, resistance to pressure compaction, wider operating temperature range and pH range, and higher stability to biological attack [5]. A major limitation that is addressed in this study is the relatively low rejection of small uncharged anthropogenic or naturally occurring molecules such as boric acid. Sea water typically contains 4–7 ppm boron (B) as boric acid [6,7], but its concentrations can also be as high as 9.6 ppm [8]. Its toxicity to humans is low therefore the World Health Organization sets a relatively high guideline value of 2.4 ppm B in drinking water [9]. However, boron becomes toxic for many crops above 0.5–1 ppm [10,11], therefore, the recommended B concentration for the desalination permeates used for irrigation is 0.3–0.5 ppm [12]. Boric acid is a weak acid with pKa1 ≈ 8.6–9.2 in seawater [13]. Since most feeds used in desalination have pH 7.5–8 [14], boric acid in seawater is mainly uncharged. RO membranes do not remove uncharged species as efficiently as salt, since ion exclusion mechanisms involved in removal of salts become inactive. Boric acid removal is then controlled mainly by size exclusion and hindrance which are not strong because of the small size of the molecule [6,15]. TFC polyamide RO membranes were shown to contain network pores having radii of about

Corresponding author. E-mail address: [email protected] (V. Freger).

http://dx.doi.org/10.1016/j.memsci.2017.10.003 Received 26 November 2016; Received in revised form 12 September 2017; Accepted 2 October 2017 Available online 06 October 2017 0376-7388/ © 2017 Elsevier B.V. All rights reserved.


S. Shultz et al.

1–3 Å and larger aggregate pores, having radii of about 3.5–4.5 Å [16–18]. Uncharged boric acid has a molecular diameter of 2.75 Å and its Stokes radius is only about 1.55 Å [19], thus it is significantly smaller than hydrated sodium (3.58 Å) and chloride (3.32 Å) ions. Boric acid size is then close to that of the network pores and substantially less than that of the aggregate pores. Boric acid can then fit in even some network pores and readily penetrate aggregate pores of RO membranes. In addition, as pointed out by Sagiv and Semiat [20], hydrogen bonding between the three hydroxyl groups of B(OH)3 and water within in the membrane may enhance association and drag by water. Due to the above factors combined sea and brackish water RO polyamide membranes have B rejection of 80–93% and 30–80%, respectively, compared ≫ 99% salt rejection [6–8,11]. A single-pass RO process is then unable to remove boron down to the required level, which can only be achieved by adding various pre- and post-treatment steps [21] or by a multi-pass RO process [22]. This increases the energy consumption and costs substantially, by 10–20% [20]. Several patents were published that target increasing the B removal in RO process; these consider the use of ion-exchangers [23], complexation with sugar derivatives [24] and membrane modification [25]. The latter has been extensively explored as a general method for improving membrane performance, as part of manufacturing or as a posttreatment. Various chemical and physical post-treatment techniques to modify membrane properties have been developed, not unique to RO membranes, aimed at improving a specific aspect of membrane performance while preserving the other advantageous characteristics [26–31]. The present study focuses on improving B rejection by modifying the polyamide layer. The approach builds upon a previous study by Bernstein et al. [32], which analyzed physical characteristics potentially beneficial for B rejection. It identified two principal ways to reduce membrane affinity towards B and thus increase its rejection: (1) reducing polarity, i.e., increasing hydrophobicity of the selective layer and (2) strengthening steric exclusion, i.e., reducing pore size. Hydrophobicity was also found in the past to be an important factor controlling the salt rejection and permeability-selectivity tradeoff [33]. In the context of B rejection, hydrophobic surfaces are known to disrupt hydrogen bonding in water, cf. the structure-breaking effect of large hydrophobic ions [34]. The role of steric exclusion, i.e., correlation of B rejection with pore size, was explicitly demonstrated by Kurihara et al. [35] and Fujioka et al. [36] using positron annihilation life-time spectroscopy (PALS). It must be noted however that interpretation of this effect as purely steric may be an oversimplification, since molecular interactions are also enhanced in smaller pores. The above principles were explored in our previous studies, in which a tight hydrophobic layer was grafted on top of polyamide, indeed improving the B rejection and permeability-rejection tradeoff, presumably by “plugging” or “caulking” largest defects in polyamide layer [37,38]. The present study explores another way to utilize the above principles through the use of small molecules that would permanently “plug” the polyamide nanopores. This should produce a double effect, namely, reduce the effective pore size of the polyamide and, simultaneously, modify its pore chemistry and polarity in a way beneficial for B rejection. The proposed change is then different from previously used surface grafting [37,38] or tightening polyamide by varying IP reaction conditions, which control effective pore size of polyamide itself, i.e., changes its structure or pore size distribution, but do not affect its inner chemistry. An important additional requirement is that the plug molecules need to be immobilized within the polyamide matrix. The rational choice of suitable molecules along these lines will be first discussed and thereafter experimental results presented that will demonstrate feasibility of the proposed approach.

2. Experimental 2.1. Materials All chemicals were purchased from Sigma-Aldrich and Acros and used without purification. Double-distilled deionized water (DDW) was used in all experiments. The fully aromatic polyamide (PA) membranes manufactured by Hydranautics, namely, SWC5 max (available in the beginning of the study), and SWC4B (used at later stages) were kindly supplied by the manufacturers as flat sheets and stored as described previously [32,38,39]. Before modification all membranes were first soaked in 50% ethanol and then in water to ensure complete pore wetting. It was verified that such wetting did not affect the initial membrane performance. 2.2. Membrane testing The filtration tests were performed in 150 mL nitrogen-pressurized dead-end stirred cells having a membrane area of 11.3 cm2. The membranes were tested at feed pressure 55 bar for water permeability (Lp) using DDW (by collecting and weighing the permeate) and for rejection of boric acid and NaCl using a solution of 32,000 ppm NaCl and 5 ppm B at pH 7–7.3. NaCl concentration in the feed and permeate was determined from electric conductance. B concentration was measured using inductively coupled plasma optical emission spectroscope (ICPOES, iCAP 6000 Series, Thermo Scientific), detection limit was 0.02 ppm. The passage (P) and rejection (1 – P) of salt and boric acid were calculated using the relation

P = Cp/ Cf ,

(1)

where Cp and Cf are the permeate and feed concentrations, respectively. Membranes that showed initial NaCl rejection below 95% were discarded. 2.3. Modification procedure Table 1 lists the molecules tested and used in the modification experiments and their main physical characteristics. An RO membrane Table 1 List of molecules used in the modifications and their characteristics. Name/ Designation

n-Alkylamines (linear) Propylamine/C3 Butylamine/C4 Amylamine/C5 Hexylamine/C6 Decylamine/C10 Dodecylamine/ C12 Alkylamines (branched) 2-(Methylbutyl) amine/C5* tert-Octylamine/ C8* Phenethylamine (aromatic)

Mw [g/ mole]

MPA [Å2]a

Volume [Å3]a

LogPa

pKab

Solubility in water [mM]c

59.1 73.1 87.2 101.2 157.3 185.4

18.1 19.8 20.1 21.4 26.5 26.9

74.1 91.1 108 125 193 227

0.25 0.70 1.14 1.59 3.37 4.25

10.21 10.21 10.6 10.56 10.64 10.63

∞ ∞ ∞ ∞ 3.5 0.42

87.2

26.51

108

1.06

10.23

∞

129.2

31.8

160

1.98

10.68

67

121.2

25.9

127

1.39

9.79

∞

a http://www.chemicalize.org, MPA is the minimal projected area, logP – is octanolwater partitioning coefficient. b https://www.ncbi.nlm.nih.gov/pccompound for amines C3, C4, C5*, C5, C6, C8*, C10, C12. c ECOSTAR- Ecological Structure Activity Relationships (ECOSAR) Predictive Model https://www.epa.gov/tsca-screening-tools/ecological-structure-activity-relationshipsecosar-predictive-model.

166


S. Shultz et al.

insertion should as well be primarily controlled by solute width or MPA. Based on the above considerations, we propose linear aliphatic amines as suitable molecular plugs meeting the above requirements. The linear amines, even longer ones, have a MPA of the order 20–25 Å2 (~ 0.3–0.5 nm width), weakly dependent on the length (Table 1), while aliphaticity renders them less polar, i.e., less polarizable and more hydrophobic, than the aromatic polyamide matrix. They may then closely fit in aggregate pores, turning them into narrower and less polar channels. For the same reason branched aliphatic or aromatic amines seem to be less suitable, since their MPA must be larger and may not fit in the pores. Higher polarizability of aromatic amines makes them also less likely to induce a desired change in the aromatic polyamide matrix. Nevertheless, an explicit comparison between linear and branched aliphatic and aromatic amines will be made below. The amine end-group seems to offer facility and flexibility of linking between the plug molecule and the matrix through unreacted carboxylic groups of polyamide. The linkage may be either ionic or covalent. The former is essentially formation of an ion pair, which should be stable in the intermediate pH range, where both carboxylic (pKa ~ 4–5) and amine groups (cf. pKa values in Table 1) used here are ionized. Alternatively, these groups may form a covalent amide bond, resulting in a more stable linkage; however, amide bond formation requires either heating or a catalyst. Carbodiimides such as EDC are well-known reagents that catalyze formation of amide bonds (see Fig. S1 in Supporting information) and are often used to link amines to molecules via carboxylic groups [47]. They were used for chemical grafting of molecules to polyamide membranes as well [48–50]. In addition to ionic or covalent linkages, hydrophobic interactions between the aliphatic tail and polyamide provide another efficient means of immobilization, especially, for larger amines. Ben David et al. showed that sorption of aliphatic n-alcohols rapidly increases with increasing aliphatic tail length, well correlated with falling solubility in water [51]. A similar effect should occur for aliphatic amines, as their solubility in water rapidly decreases and logP value, indicative of the partitioning between polyamide and water [51], increase when aliphatic tail length exceeds 5 carbons (Table 1). Note Table 1 lists solubility of free molecules, above which largest amines may still stay in solution as micelles, yet micellized amine fraction is not expected to affect amine uptake by polyamide, since the uptake is controlled by the free molecule concentration. Enhanced uptake of hydrophobic amines may also facilitate the amidation reaction in the membrane and strengthen ionic linkage, since electrostatic attraction is enhanced in less polar media. In either case, a synergy between uptake and immobilization mechanisms may be expected for longer aliphatic amines. The treatment with amine solutions as modifying agents was proposed in patent literature some 3 decades ago, however, the use of amines longer than 7 carbons was ruled out as resulting in a reduced flux [25,52]. Nevertheless, the benefits of a larger change in the intrinsic selectivity offered by larger amines could offset the drop in permeability, especially, in the context of boron removal. To some degree, lower permeability may be avoided by modifying a more permeable, i.e., thinner membrane, yet this option obviously requires some care, as thinner membranes usually have a lower salt rejection, which may be insufficiently improved by modification. The ultimate benefits are best judged by considering permeability-selectivity tradeoff, as discussed in Section 3.3.

(SWC5 max or SWC4B, kindly supplied by Hydranautics) was mounted in a dead-end cell and initial performance was tested. Thereafter, the membrane was modified using one of two alternative procedures, coupling (reaction) or sorption. In the coupling procedure the cell was filled with a solution of 0.1 g of coupling agent N-(3-Dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (EDC, see Fig. S1 in Supplementary materials) to activate carboxylic groups for 3 h at temperature between 0 and 4 C degrees at pH of 4.6–4.8. At the end of the activation stage, without disassembling the cell, the membrane was washed several times with DDW and the cell was filled with the solution of a selected amine molecule to react for 24 h at temperature 0–4 °C. Finally, the membrane was washed thoroughly with DDW and left in water overnight before testing for permeability and B and NaCl rejection. In the sorption procedure, after testing the initial membrane performance, the dead-end cell was filled with a solution of the desired molecule and the solution was filtered for about an hour at pressure 55 bar for prescribed time. Finally, the modifying solution was removed and the system was washed with DDW without disassembling the cell, left in water overnight, and then the membrane performance was tested again. 2.4. Surface characterization Attenuated total reflection (ATR) FTIR spectra (average of 40 scans at 4 cm−1 resolution) were recorded on a Nicolet 8700 FTIR spectrometer (Thermo- Electron corporation) using a Miracle ATR attachment with a one-reflection diamond element (Pike). The contact angle was measured by the sessile drop method using a Kruss contact angle analyzer. Every measurement was repeated and averaged for at least 7 drops on each membrane sample. Zeta potential of pristine and dodecylamine-modified membranes was measured using SurPASS with a variable gap cell (Anton Paar) in 10 mM KCl solution; each measurement was repeated 2–3 times. Surface morphology of SW and BW membranes was examined using scanning electron microscope (Zeiss Ultra-Plus, Germany). Prior to imaging the membrane samples were dried in oven at 30 °C for 12 h and coated with gold. XPS measurements were conducted using ESCALAB 250 (Thermo Fisher Scientific Inc., Waltham, UK) with Al X-ray source and monochromator. 3. Results and discussion 3.1. Rationales for selection of the modifying molecules As discussed by Bernstein et al. [32,38,39] and explained in Introduction, the rejection of B may benefit from (a) size exclusion and (b) reduced polarity (i.e., molecular polarizability) of the selective layer. Our recent experiments and simulations indicate that the polyamide membrane is essentially a random porous network, in which some critical passages determine the water and solute permeation [40–42]. The route explored in this study is essentially based on tightening or blocking such passages. Specifically, in the polyamide pores, small hydrophobic molecules may act as molecular plugs that change both the polyamide structure and chemistry. Such plug molecules must be immobilized within the pores to ensure a sustainable change in performance. This may be realized in two principal ways: through strong interactions with the polyamide matrix or via covalent bonding. To maximize the insertion and plugging effect, additional factors may be considered. Published reports on filtration of molecular solutes (usually, pollutants) indicate that membrane permeability may drop significantly when the solutes either have a high affinity hence are strongly uptaken by the membrane [43,44] or have a size closely fitting that of the membrane pores [45]. Several authors pointed out parameters such as molecular width [46] or minimal projected area (MPA) [36] best correlate with passage through membranes. This suggests that

3.2. Modification by amidation reaction In the first series of experiment amidation coupling procedure was used, targeting formation of a covalent amide bond between the amine and carboxylic groups of polyamide catalyzed with carbodiimide. The original membrane was either SWC5max used for modifying C3, C4, C5*, C5, C6, C8* amines or SWC4B for C10, C12 amines. (The designation indicates the number of carbons in amine and a star means that aliphatic tail is branched – see Table 1). The two SWRO membranes had 167


S. Shultz et al.

Fig. 1. Membrane performance before and after modification with different amine molecule via EDCmediated amidation. The molecules propylamine (C3), butylamine (C4), 2(methylbutyl)amine (C5*), and tert-octylamine (C9*) were used at 15.64 mM concentration, molecules amylamine (C5), hexylamine (C6), decylamine (C10) and dodecylamine (C12) were used at 1.04 mM concentration. The original membrane was SWC5max for amidation with C3, C4, C5*, C5, C6, C8* amines and SWC4B for amidation with C10 and C12. (A) Water volume flux for pure water at 55 bar; the shaded areas correspond to the range of measured water permeability for respective pristine membranes, the width corresponds to standard deviation, (B) boric acid passage, (C) salt passage, (D) B passage versus MPA for different amine molecules The dashed lines in (B) and (D) indicate the trend for linear n-alkylamines. Testing conditions for (B) to (D): [NaCl] = 32,000 ppm, [B] = 5 ppm, pressure 55 bar. Each result represents at least 4 repetitions.

n-alkylamine trend SWC4B

A

B

9 8 Salt passage[%]

7

Before

C8 *

C4

After

C3

6

C6

5

C5* C10

C5

4 3

C12

2 1 0 C3 C4 C5* C5 C6 C8* C10 C12

C

D resulting in much scatter of the results both before and after modification. The individual results are then less indicative than the general trend, which indicates that for linear n-alkylamines the B passage steadily improves as the amine length increases and becomes very substantial for C10 and C12. The NaCl rejection also showed some improvement, relative to the initial performance of the same samples before modification, though less significant (Fig. 1C). The trend is consistent with the assumptions discussed in Section 3.1. Fig. 1D, plotting B rejection of modified membranes versus MPA of the modifying amine molecule, shows that branched tert-octylamine (C8*) falls out of the trend and produces little change in B rejection. This could be anticipated, based on its much larger MPA (Table 1 and Fig. 1D), which could prevent its penetration to the smallest nanopores, in which linear amines could still fit. On the other hand, another branched molecule, 2-(Methylbutyl) amine C5*, fell close to the trend, probably, since its size and width, similar to linear amines, could still allow a reasonable insertion in the small pores. Similar tests performed with aromatic phenethylamine (not shown) did not lead to any improvement in performance and NaCl and boron rejection even worsened, which was anticipated (see previous section). Moreover, exposure to higher concentrations of this amine visibly damaged the polysulfone layer therefore no further tests with this molecule were carried out. ATR-FTIR could be used to directly detect the uptake of aliphatic amines, which show up as aliphatic C-H stretching bands emerging in the 2800–3000 cm−1 region (Fig. 2), where aromatic polyamide, whose structure is shown in Fig. 2 as well, and polysulfone support have no or weak IR absorption. In other spectral regions, the spectrum was not affected, suggesting the membrane was not damaged or chemically altered. Unfortunately, the intensity of the emerging aliphatic bands showed large variations and did not correlate well with changes in

an identical chemistry and similar permeability and differed mainly in the nominal B rejection (Table S1 in Supplementary materials), which was fairly insignificant, given large variations in performance measured for small samples used in dead-end cell tests. These experiments were performed in two steps: first activation of carboxylic groups with EDC (see Fig. S1 in Supplementary materials) and thereafter exposure to an amine solution. The amine concentration in solution turned out to be an important factor that required adjustment according to amine molecular weight and solubility in water (Table 1). Some drop in water permeability relative to initial performance was observed in all cases (Fig. 1A). The preliminary experiment with 50.2 mM C5* solution, a relatively small branched amine, already showed an excessive drop in permeability therefore in subsequent experiments the amine concentration was reduced to 15.64 mM. This resulted in an acceptable permeability after modification for shorter linear amines, C3 to C6, as well as for branched C8*, however, for the longer linear amines, C10 and C12, the permeability was affected severely and dropped to a very low, near-zero value. Apparently, excessive uptake of amine rendered the membrane too hydrophobic and virtually impermeable to water. It is likely that amine micelles fouled the surface as well, since the C10 and C12 concentration was above the free molecule solubility. To mitigate the loss of permeability, in subsequent experiments the C10 and C12 concentration was further reduced to 1.04 mM, below the free molecule solubility of C10 and only slightly above it for C12. This resulted in an acceptable drop in permeability, therefore the subsequent measurements used 15.64 mM amine concentration for C3, C4, C5* and C8* amines and 1.04 mM for C5 and longer linear amines. Fig. 1B compares B passage before and after modification for different amines. Unfortunately, the initial performance of small membrane samples used in these experiments showed significant variations 168


S. Shultz et al.

Fig. 2. ATR-FTIR spectra of a pristine polyamide RO membrane and membranes modified using EDC-mediated amidation with propylamine and dodecylamine. Inset shows the polyamide structure with reactive groups encircled.

Dodecylamine Aliphatic C-H stretching bands

Propylamine

Carbonyl band1,720 cm-1

Pristine

with the major IR bands of polyamide. Some amine could also be bound non-covalently, due to affinity to polyamide or ionic interactions. The solubility and logP values listed in Table 1 suggest that the uptake of amines could differ by as much as 3–4 order of magnitude; thereby the physically sorbed fraction could eventually surpass the covalently bonded one for the longer amines. To analyze this point, another series of modification experiments was performed without activation step by simple filtration of the amine solution, in which only physical sorption or ionic binding of amines could occur. The change in performance of the membranes modified using the sorption procedure is summarized in Fig. 3. Note the pure water flux of the SWC4B membrane samples in these tests was found to be somewhat higher and initial B passage somewhat lower than in coupling experiments (cf. Fig. 1). Once again, this emphasizes inherent variability of flat sheet batches and small samples, making the overall trends and relative (pair-wise) differences before and after modification more indicative than individual results. Clearly, C5 and C6 sorption did not significantly affect or even slightly worsened the B passage (Fig. 3B), in contrast to the respective results for coupling (Fig. 1B). However, simple sorption of C10 and C12 improved the B passage to a degree commensurate with coupling procedure. The results suggest that, in general, the carbodiimide-assisted procedure could result in at least some covalent bonding, leading to an improvement in selectivity. However, for longer amines the sorption contribution to improving B rejection seems to become increasingly important and, likely, dominant. The water passage shows little change after sorption of shorter amines; however a more significant flux reduction (Fig. 3A) occurs for the longest aliphatic amines, similar to the results for coupling (Fig. 1A). Fig. 3C indicates that for all tested amines there was no significant change or even slight improvement of salt passage after amine sorption. ATR-FTIR was also used to verify immobilization of amine after sorption. It showed that the aliphatic bands of the amine, similar to those seen in Fig. 2, sustained after 48 h soaking in water at 60 °C with water exchanged several times. This suggests that after sorption the longer amines were strongly and sustainably immobilized within polyamide. Curiously, the contact angle did not change after sorption as much as after amidation or even slightly decreased. Probably, as in surfactant sorption, it was the hydrophobic tail rather than amine groups that adhered to the membrane with the (non-bound) hydrophilic amine groups facing outward, producing no or little overall change in water contact angle. Finally, streaming potential measurements of pristine and C12-treated membranes showed no virtually no difference (Fig. S3 in Supplementary materials). The surface characteristics along with ATR-FTIR data suggest that the change produced by amine

performance and amine length. Presumably, this reflected variations in polyamide thickness and nanostructure (hence in the amount of amine that could be uptaken) and could only be used as a semi-quantitative indicator. ATR-FTIR was also used to verify the strong immobilization of amine after the modification. This confirmed that the change seen in ATR-FTIR spectra sustained and modified performance remained stable even after 48 h of soaking in water at 60 °C with water exchanged several times to simulate extended exposure at ambient conditions. These results suggest that amine was strongly immobilized within polyamide. Interestingly, in modified membranes a small band emerged also at 1720 cm−1, assigned to the carbonyl of the carboxylic group (encircled in the polyamide structure in Fig. 2). It could be speculated that amine uptake might result in hydrolysis of a small fraction of the amide bonds, which could in turn contribute to immobilization of the amine. Water contact angles measurements offer another insight. The method obviously cannot provide information on changes within polyamide, but can shed light on amine binding. The untreated SWC4B and SWC5max membranes showed a similar water contact angles of 42.0 ± 2.4° and 46.6 ± 7.6°, respectively, whereas the water contact angle of SWC4B membranes treated with decylamine and dodecylamine increased slightly but significantly to 54.3 ± 1.9° and 51.1 ± 2.3°. These results could indicate that binding of long aliphatic amines to an EDC-activated membrane resulted in a slightly more hydrophobic surface. Possibly, when amine group is covalently or ionically anchored, the hydrophobic aliphatic moieties could tend to face outwards, making a more hydrophobic surface. In any case amine layer remains extremely thin, as SEM showed no change in surface morphological after coupling treatment with C12 (see Fig. S2 in Supplementary materials). In addition, XPS measurements, probing a few nanometers thick surface layer, showed no significant variations of the surface atomic composition. For instance, the O content after coupling with C10 and C12 stayed fairly close to that of the pristine membrane, 14–15% (see Table S2 in Supplementary materials), consistent with other data for polyamide membranes [35], while already for a 10 nm thick amine layer, i.e., well above penetration depth of XPS, it would drop close to 0. 3.3. Modification by amine sorption Although the above coupling experiments were supposed to result in formation of amide linkages, it was not clear how much amine was actually covalently bound via the amide bonds to polyamide matrix. The amount of such bonds was too small to produce a consistent and detectable change in ATR-FTIR spectra, given the overlap of such bands 169


S. Shultz et al.

40

Pristine membranes

Before After

35

100

30

B passage [%]

Water flux [L/h*m2]

120

80 60 40 20

25 20

Fig. 3. Performance of SWC4B membrane before and after modification by sorption of amylamine (C5), hexylamine (C6), decylamine (C10), and dodecylamine (C12) from 1.04 mM solution. (A) Pure water flux at 55 bar; dashed line indicates average value for the pristine membrane, (B) boric acid passage and (C) salt passage. Testing conditions: [B] = 5 ppm, [NaCl] = 32,000 ppm, pressure 55 bar. Each bar represents average of at least 4 samples.

15 10 5 0

0 C5

C6

C10

C5

C12

C6

C10

C12

B

A

Salt passage [%]

7 6

Before

5

After

4 3 2 1 0 C5

C6

C10

C12

C sorption was mainly within the polyamide and not on the surface. 3.4. Effect of modification on permeability-selectivity tradeoff RO membranes usually show a tradeoff between selectivity and permeability, i.e., the B rejection generally improves, as membrane is made tighter and less permeable. This tradeoff may be viewed as an intrinsic characteristic of the selective material, i.e., polyamide. A successful modification is supposed to move the trend downward, when solute passage is plotted versus permeability, as in Fig. 4. Note that the commercial membranes usually show a much better selectivity in spiral-wound elements than in small flat-sheet coupons in dead-end cells employed in this study. Specifically, the B passage of SWRO membranes as commercial spiral-wound elements is usually 8–10%, while in dead-end cells it was 30–45%. The difference is believed to be due to different hydrodynamic and concentration polarization conditions in the cells, possible differences in membrane post-treatment and damage or defects sustained by small flat sheet samples during handling and mounting in the cells. Thus, elements and small flat sheet samples tested in laboratory cells cannot be compared. Therefore, for consistency, we compare below the pristine and modified membranes tested in the dead-end cells in the same conditions. Fig. 4 summarizes the trade-off obtained for all membranes, pristine and modified. Note that, due to availability at the time of experiments, the starting membranes were not identical for short amines in amidation and sorption procedures, however, the two membranes used, SWC4B and SWC5max, differed only moderated in permeability and B rejection. It is seen that modification either kept the membrane on the same trend as the original membranes or consistently improved the tradeoff compared to the original one. This indicates that modified

Fig. 4. Permeability-selectivity trade-off plot: boric acid passage versus membrane permeability. Diamonds correspond to pristine SWC5max and SWC4B membranes tested in dead-end cells, circles to membranes modified via carbodiimide-assisted coupling, and triangles to modification by sorption. The original membrane was SWC5max for amidation with C3, C4, C5*, C5, C6, C8* and SWC4B for amidation with C10 and C12 and for all molecules in sorption procedure. Performance testing conditions: [B] = 5 ppm, [NaCl] = 32,000 ppm, pressure 55 bar. Each symbol represents the average of at least 4 results.

170


S. Shultz et al.

[16] S.H. Kim, S.-Y. Kwak, T. Suzuki, Positron annihilation spectroscopic evidence to demonstrate the flux-enhancement mechanism in morphology-controlled thin-filmcomposite (TFC) membrane, Environ. Sci. Technol. 39 (2005) 1764–1770. [17] O. Coronell, B.J. Mariñas, X. Zhang, D.G. Cahill, Quantification of functional groups and modeling of their ionization behavior in the active layer of FT30 reverse osmosis membrane, Environ. Sci. Technol. 42 (2008) 5260–5266. [18] D.G. Cahill, V. Freger, S.-Y. Kwak, Microscopy and microanalysis of reverse-osmosis and nanofiltration membranes, MRS Bull. 33 (2008) 27–32. [19] J.K. Park, K.J. Lee, Diffusion coefficients for aqueous boric acid, J. Chem. Eng. Data. 39 (1994) 891–894. [20] A. Sagiv, R. Semiat, Analysis of parameters affecting boron permeation through reverse osmosis membranes, J. Membr. Sci. 243 (2004) 79–87. [21] Y. Xu, J.Q. Jiang, Technologies for boron removal, Ind. Eng. Chem. Res. 47 (2008) 16–24. [22] M. Faigon, D. Hefer, Boron rejection in SWRO at high pH conditions versus cascade design, Desalination 223 (2008) 10–16. [23] J. McMullen, W. Tsu, R. Kargel, Removal of Boron and Fluoride from Water, U.S. Patent No. 6,296,773, 2 October 2001. [24] K. Yabusaki, Boron Removal Method Utilizing Sugar Amide Derivative, U.S. Patent No. 8,236,180. 7 August 2012. [25] H. Tomioka, K. Sugita, K. Oto, Composite Semipermeable Membrane, and Production Process Thereof, U.S. Patent No. 7,279,097. 9 October 2007. [26] M. Ulbricht, Advanced functional polymer membranes, Polymer 47 (2006) 2217–2262. [27] D. Rana, T. Matsuura, Surface modifications for antifouling membranes, Chem. Rev. 110 (2010) 2448–2471. [28] K.C. Khulbe, C. Feng, T. Matsuura, The art of surface modification of synthetic polymeric membranes, J. Appl. Polym. Sci. 115 (2010) 855–895. [29] I. Pinnau, B.D. Freeman, Formation and modification of polymeric membranes: overview, in: I. Pinnau, B.D. Freeman (Eds.), Membrane Formation and Modification, ACS Symposium Series, Vol 744, Am. Chem. Soc. 1999, pp. 1–22. [30] K. Kato, Polymer surface with graft chains, Prog. Polym. Sci. 28 (2003) 209–259. [31] A.L. Zydney, L. Zeman, Microfiltration and Ultrafiltration: Principles and Applications, Marcel Dekker, New York, NY, 1996. [32] R. Bernstein, S. Belfer, V. Freger, Toward improved boron removal in RO by membrane modification: feasibility and challenges, Environ. Sci. Technol. 45 (2011) 3613–3620. [33] H. Strathmann, A.S. Michaels, Polymer-water interaction and its relation to reverse osmosis desalination efficiency, Desalination 21 (1977) 195–202. [34] R.A. Robinson, R.H. Stokes, Electrolyte Solutions, 2nd revised ed., Butterworths, London, 1965. [35] M. Kurihara, T. Sasaki, K. Nakatsuji, M. Kimura, M. Henmi, Low pressure SWRO membrane for desalination in the Mega-ton water system, Desalination 368 (2015) 135–139. [36] T. Fujioka, N. Oshima, R. Suzuki, W.E. Price, L.D. Nghiem, Probing the internal structure of reverse osmosis membranes by positron annihilation spectroscopy: gaining more insight into the transport of water and small solutes, J. Membr. Sci. 486 (2015) 106–118. [37] A. Ben-David, R. Bernstein, Y. Oren, S. Belfer, C. Dosoretz, V. Freger, Facile surface modification of nanofiltration membranes to target the removal of endocrine-disrupting compounds, J. Membr. Sci. 357 (2010) 152–159. [38] R. Bernstein, S. Belfer, V. Freger, Surface modification of dense membranes using radical graft polymerization enhanced by monomer filtration, Langmuir 26 (2010) 12358–12365. [39] R. Bernstein, S. Belfer, V. Freger, Improving performance of spiral wound RO elements by in situ concentration polarization-enhanced radical graft polymerization, J. Membr. Sci. 405–406 (2012) 79–84. [40] E. Dražević, K. Košutić, V. Freger, Permeability and selectivity of reverse osmosis membranes: correlation to swelling revisited, Water Res. 49 (2014) 444–452. [41] V. Kolev, V. Freger, Hydration, porosity and water dynamics in the polyamide layer of reverse osmosis membranes: a molecular dynamics study, Polymer 55 (2014) 1420–1426. [42] V. Kolev, V. Freger, Molecular dynamics investigation of ion sorption and permeation in desalination membranes, J. Phys. Chem. B 119 (2015) 14168–14179. [43] M.E. Williams, J.A. Hestekin, C.N. Smothers, D. Bhattacharyya, Separation of organic pollutants by reverse osmosis and nanofiltration membranes: mathematical models and experimental verification, Ind. Eng. Chem. Res. 38 (1999) 3683–3695. [44] E. Drazevic, S. Bason, K. Kosutic, V. Freger, Enhanced partitioning and transport of phenolic micropollutants within polyamide composite membranes, Environ. Sci. Technol. 46 (2012) 3377–3383. [45] B. Van Der Bruggen, C. Vandecasteele, Flux decline during nanofiltration of organic components in aqueous solution, Environ. Sci. Technol. 35 (2001) 3535–3540. [46] Y. Kiso, Effects of hydrophobicity and molecular size on rejection of aromatic pesticides with nanofiltration membranes, J. Membr. Sci. 192 (2001) 1–10. [47] A. Okamura, T. Hirai, M. Tanihara, T. Yamaoka, Synthesis and properties of novel biodegradable polyamides containing α-amino acids, Polymer 43 (2002) 3549–3554. [48] A.-F. Che, F.-Q. Nie, X.-D. Huang, Z.-K. Xu, K. Yao, Acrylonitrile-based copolymer membranes containing reactive groups: surface modification by the immobilization of biomacromolecules, Polymer 46 (2005) 11060–11065. [49] M. Li, Z. Lv, J. Zheng, J. Hu, C. Jiang, M. Ueda, X. Zhang, L. Wang, Positively charged nanofiltration membrane with dendritic surface for toxic element removal, ACS Sustain. Chem. Eng. (2016) 784–792. [50] A.M.S. de Jubera, J.H. Herbison, Y. Komaki, M.J. Plewa, J.S. Moore, D.G. Cahill, B.J. Marinas, Development and performance characterization of a polyannide

membranes display a more favorable combination of performance characteristics than original commercial membranes. The improvement was largest for the longest amines and slightly better for EDC-mediated procedure. However, for longest C10 and C12 amine the facile sorption procedure may nearly as beneficial as the more tedious amidation. 4. Conclusions The study examined the possibility to improve boron rejection of commercial see water RO membranes using reactive coupling or sorption of simple aliphatic amine molecules from solutions. The present results are viewed mainly a feasibility test and the ultimate conclusions will require testing the procedure on real elements, which is reported elsewhere and largely confirm present conclusions [53]. Clearly, much optimization and, possibly, search for better plug molecule along proposed rationales, may lead to further improvement. Nevertheless, the present results demonstrate that tailoring performance using in-situ modification with small “molecular plugs” is feasible. The carbodiimide-based chemistry showed potential for achieving best results and this or analogous coupling chemistry may be used during membrane manufacturing to tune and maximize the selectivity and overall performance. However, the facile sorption procedure, requiring simple exposure to dilute amine solutions and nearly as efficient for longer amines as coupling, could be useful for in situ modification of commercial elements and even existing installations as well. Curiously, the proposed treatment may be seen as a kind of fouling and pore-blocking has long been known as one of the fouling mechanisms [54]. However, while fouling and biofouling are usually detrimental to performance and reduce boron rejection (e.g., [55]), in the present case “engineered fouling” becomes beneficial for the performance and, especially, membrane selectivity. Appendix A. Supporting information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.memsci.2017.10.003. References [1] M.E. Williams, A Brief Review of Reverse Osmosis Membrane Technology, EET Corporation and Williams Engineering Services Company, Harriman, TN, 2003. [2] L.F. Greenlee, D.F. Lawler, B.D. Freeman, B. Marrot, P. Moulin, Reverse osmosis desalination: water sources, technology, and today's challenges, Water Res. 43 (2009) 2317–2348. [3] T Uemura, K Kotera, M Henmi, H Tomioka, Membrane technology in seawater desalination: history, recent developments and future prospects, Desalin. Water Treat. 33 (1-3) (2012) 283–288. [4] R.J. Petersen, Composite reverse osmosis and nanofiltration membranes, J. Membr. Sci. 83 (1993) 81–150. [5] D. Li, H. Wang, Recent developments in reverse osmosis desalination membranes, J. Mater. Chem. 20 (2010) 4551–4566. [6] V. Freger, H. Shemer, A. Sagiv, R. Semiat, Boron removal using membranes, in: N. Hilal, E. Kabay, M. Bryjak (Eds.), Boron Separation Processes, Elsevier, 2015, pp. 199–219. [7] M. Busch, W.E. Mickols, S. Jons, J. Redondo, J. De Witte, Boron removal in sea water desalination, Int. Desalin. Water Reuse Q. 13 (2004) 25. [8] K.L. Tu, L.D. Nghiem, A.R. Chivas, Boron removal by reverse osmosis membranes in seawater desalination applications, Sep. Purif. Technol. 75 (2010) 87–101. [9] World Health Organization, Boron in Drinking-Water: Background Document for Development of WHO Guidelines for Drinking-Water Quality, World Health Organization, Geneva, 2009. [10] R.O. Nable, G.S. Bañuelos, J.G. Paull, Boron toxicity, Plant Soil. 193 (1997) 181–198. [11] N. Hilal, G.J. Kim, C. Somerfield, Boron removal from saline water: a comprehensive review, Desalination 273 (2011) 23–35. [12] P. Glueckstern, M. Priel, Optimization of boron removal in old and new SWRO systems, Desalination 156 (2003) 219–228. [13] I. Hansson, A new set of acidity constants for carbonic acid and boric acid in sea water, Deep Sea Res. Oceanogr. Abstr. 20 (1973) 461–478. [14] J. Redondo, M. Busch, J.P. De Witte, Boron removal from seawater using FILMTEC TM high rejection SWRO membranes, Desalination 156 (2003) 229–238. [15] H. Koseoglu, N. Kabay, M. Yüksel, S. Sarp, Ö. Arar, M. Kitis, Boron removal from seawater using high rejection SWRO membranes – impact of pH, feed concentration, pressure, and cross-flow velocity, Desalination 227 (2008) 253–263.

171


S. Shultz et al.

[53] S. Shultz, V. Freger, In situ modification of membrane elements for improved boron rejection in RO desalination, Desalination (2017), http://dx.doi.org/10.1016/j. desal.2017.08.021. [54] R. Field, Fundamentals of fouling, in: K.-V. Peinemann, S.P. Nunes (Eds.), Membranes for Water Treatment, v. 4, Wiley-VCH, Weinheim, 2010. [55] E. Huertas, M. Herzberg, G. Oron, M. Elimelech, Influence of biofouling on boron removal by nanofiltration and reverse osmosis membranes, J. Membr. Sci. 318 (2008) 264–270.

nanofiltration membrane modified with covalently bonded aramide dendrimers, Environ. Sci. Technol. 47 (2013) 8642–8649. [51] A. Ben-David, Y. Oren, V. Freger, Thermodynamic factors in partitioning and rejection of organic compounds by polyamide composite membranes, Environ. Sci. Technol. 40 (2006) 7023–7028. [52] W.E. Mickols, Z. Chunming, Polyamide Membrane with Coating of Polyalkylene Oxide and Polyacrylamide Compounds, U.S. Patent No. 7,815,987, 19 October 2010.

172