Structure-Motion-Performance Relationship of Flux-Enhanced

Sep 27, 2001 - Most of the commercially successful RO TFC membranes are FT-30 ... The DMSO was added in the aqueous solution in the three different co...
0 downloads 0 Views 255KB Size
Download PDF
Environ. Sci. Technol. 2001, 35, 4334-4340

Structure-Motion-Performance Relationship of Flux-Enhanced Reverse Osmosis (RO) Membranes Composed of Aromatic Polyamide Thin Films SEUNG-YEOP KWAK,* SOO GYUNG JUNG, AND SUNG HO KIM Hyperstructured Organic Materials Research Center (HOMRC), and School of Materials Science and Engineering, Seoul National University, San 56-1, Shinlim-dong, Kwanak-ku, Seoul 151-744, Korea

The present paper explores the role of dimethyl sulfoxide (DMSO) used as an additive to modify the morphological as well as the molecular nature of aromatic polyamide during the formation of thin-film-composite (TFC) membranes. In addition, it elucidates the mechanism of enhancing the reverse osmosis (RO) permeation of the resulting membranes in proportion to the addition of DMSO. Morphological studies by atomic force microscopy (AFM) observed that as the concentration of DMSO increased, the surface roughness and the surface area of the aromatic polyamide TFC membranes became higher and larger, compared to FT-30 membrane for which DMSO was not added during interfacial reaction. Such morphological changes were brought about from fluctuating interface through reducing the immiscibility between aqueous/organic phases by DMSO and provided more opportunities to have contact with water molecules on the surface, participating in the enhancement of the water permeability. Chemical composition studies by X-ray photoelectron spectroscopy (XPS) revealed that there was a considerable increase of the crosslinked amide linkages relative to the linear pendant carboxylic acid groups in the TFC membranes of more DMSO addition. The increase of such amide linkages as hydrogen bonding sites facilitated the diffusion of water molecules through the thin films and played a favorable role in elevating water flux without considerable loss of salt rejection. Relaxation and motion analyses by 1H solid-state nuclear magnetic resonance (NMR) spectroscopy also confirmed the XPS revelation on the basis of measurements of the spin-lattice relaxation time in the rotating frame, T1F, and determination of the correlation time, τc, for the aromatic polyamides forming thin films. The trend of longer τc’s with the increase of DMSO concentration reflected the thin-film aromatic polyamides of less locally mobile chains, accompanied by the higher degree of cross-linking and, hence, the greater number of amide groups. The combined results of AFM, XPS, and solid-state NMR provided a robust explanation for the mechanism of flux enhancement of the aromatic polyamide TFC membranes with the addition of DMSO, which would contribute to not only a fundamental understanding of the process but also an

advanced designing of the so-called “tailor-fit” TFC membranes.

Introduction Since Cadotte (1, 2) discovered the aromatic polyamide thinfilm-composite (TFC) membranes for the reverse osmosis (RO) process, the TFC RO membranes have become indispensable in our daily life and many industrial areas such as desalting of brine, ultrapure water production, environmental pollution treatment, and so on (3-5). The trend for the next generation is to require more sophisticated and specified function of polymeric materials as well as performance. Thus, the need of so-called “tailor-fit” materials whose functions and properties are precisely tuned for the intended application is very strong. It has also become essential to tailor-fit a reverse osmosis (RO) thin-film-composite (TFC) membrane more accurately to the specific application where high performance or novel function is needed. This can be achieved with either (i) design and synthesis of totally new polymers forming thin films of the RO membranes or (ii) physical/chemical modification of the thin films (6-13). The former approach results in TFC membranes of enhancing water flux but simultaneously an accompanying and considerable loss of salt rejection or vice versa. The latter approach results from two routes that are the modification by posttreatment of the thin-film surface with various chemicals or the use of additives during the formation of the thin film. For the latter, a number of recent studies have reported that the water flux increases significantly up to an order of magnitude, maintaining a reasonable salt rejection, and in some cases both the transport characteristics are simultaneously improved. Most of the commercially successful RO TFC membranes are FT-30 type, that is the thin-film active layer which consists of aromatic polyamide prepared by the interfacial polymerization of m-phenylenediamine (MPD) in the aqueous phase and trimesoyl chloride (TMC) in the organic phase. From the viewpoint of performance efficiency, TFC membranes are usually required to have dramatically enhanced water permeability without sacrificing salt separability. Such aromatic polyamide TFC membranes with excellent water flux and reasonable salt rejection characteristics are formed by the interfacial reaction of MPD/TMC in the presence of alcohols, ethers, sulfur-containing compounds, and monohydric aromatic compounds as additives (14-16). Among the additives, dimethyl sulfoxide (DMSO) has been reported to greatly improve the water flux, and the DMSO added in the aqueous phase is known to increase the miscibility between two immiscible phases, thereby facilitating the diffusion of MPD to the organic TMC phase (16). This reaction results in modification in surface morphology and variation in the polymer chain organization forming the thin film, compared to the chemically analogous FT-30, a nonadditive version of MPD/TMC. Nevertheless, few studies have been performed to understand the behavior and mechanism of the permeability enhancement when the DMSO is used as an additive. Thus, the objective of this study is to correlate the enhanced transport performance with the changes in the surface characteristics and the molecular structure which would contribute to the fundamental understanding of the flux enhancement and eventually to the designing of new TFC membranes with advanced performance. * Corresponding author phone: +82-2-880-6082; fax: +82-2-8766086; e-mail: [email protected].

4334

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 21, 2001

10.1021/es010630g CCC: $20.00

 2001 American Chemical Society Published on Web 09/27/2001

TABLE 1. Composition of Diamine and Acid Chloride Solutions for the Formation of Aromatic Polyamide Thin Films sample

MPD

FT-30 A1 A2 A3 A4

2.0 2.0 2.0 2.0 2.0

diamine soln (wt %) DMSO TEA/CSA 0.0 1.0 2.0 3.0 4.0

1.1/2.3 1.1/2.3 1.1/2.3 1.1/2.3 1.1/2.3

acid chloride soln (wt %) TMC 0.1 0.1 0.1 0.1 0.1

Recognizing that the membrane surface governs the actual separation of the solute and the passage of the solvent, it is of prime importance to characterize the morphological variations on the surface of the TFC membranes, in conjunction with the role of DMSO. In this study, atomic force microscopy (AFM) was used as direct means to investigate and analyze the morphological characteristics of the TFC membranes. In addition to the surface morphological structure, the variation of the molecular structure in conjunction with the addition of DMSO plays a crucial role in affecting the permeability enhancement of the TFC membranes. X-ray photoelectron spectroscopy (XPS), also known as electron spectroscopy for chemical analysis (ESCA), was performed to analyze the variation of chemical composition in the thin-film polymers underlying the enhancement of the water flux. Another important factor that is inevitably mutually interdependent on the molecular structure and may affect the membrane permeability is associated with the molecular relaxation and local motion of the thin-film polymers in their solid state. The local chain mobility can be evaluated by solid-state nuclear magnetic resonance (NMR) spectroscopy. Pulsed wide-line 1H NMR relaxation analysis by measuring spin-lattice relaxation time in the rotating frame, T1F, and determining the correlation time, τc, provides a convenient way to get information about average local motion in tens of KHz frequency region for the whole thinfilm polymer.

Experimental Section Preparation of Thin-Film-Composite (TFC) Membranes and Measurement of Transport Characteristics. The thin film layer of the reverse osmosis (RO) TFC membranes is made via interfacial polymerization of m-phenylenediamine (MPD) in aqueous phase and trimesoyl chloride (TMC) in organic phase (n-hexane). Dimethyl sulfoxide (DMSO), triethylamine (TEA), and camphorsulfonic acid (CSA) were added to the amine water solution. It is noted that TEA (1.1 wt %) and CSA (2.3 wt %) form amine salts in the aqueous phase and are believed to prevent the thin film from shrinkage when drying (17). The DMSO was added in the aqueous solution in the three different compositions of 1.0, 2.0, 3.0, and 4.0 wt %, respectively (refer to Table 1). Then, the MPD solutions were coated on the nonwoven fabric-reinforced polysulfone microporous supports, excess solution of MPD was removed, TMC solution (0.1 wt %) was coated thereon, and finally the resulting thinfilm layer was dried at 95 °C for 210 s. The resulting aromatic polyamide TFC membranes are referred to as A1, A2, A3, and A4, respectively. For a comparison purpose, the bare MPD/ TMC TFC membrane, i.e., FT-30, was also prepared as a reference. Reverse osmosis (RO) performance tests were conducted at an operating pressure of 225 psi using 2000 ppm NaCl solution at 25 °C with the apparatus of a cross-flow type. Water flux was determined by direct measurement of the permeate flow:

flux (gfd) )

permeate (gallon)

(1)

membrane area (ft2)‚time (day)

Salt rejection was measured by the salt concentration in the permeate obtained through measurements of the electrical conductance of the permeate and the feed using a conductance meter (Orion model 162):

(

rejection (%) ) 1 -

permeate conductance × 100 feed conductance (2)

)

Atomic Force Microscopy (AFM). The surface topologies of the aromatic polyamide TFC membranes were investigated and analyzed with atomic force microscope (Park Scientific Instruments AutoProbe M5). In the operation of AFM, the membrane surfaces were scanned in noncontact (NC) mode, which is appropriate for imaging samples of low moduli such as soft polymers that can easily be damaged by the tip. To obtain the highest lateral resolution, images were taken with 512 × 512 pixels. Scanning rates were chosen in the range from 0.6 to 1 Hz. X-ray Photoelectron Spectroscopy (XPS). XPS experiments were performed with a Kratos AXIS HS spectrometer using Mg KR X-ray source (1253.6 eV). The X-ray gun was operated at 10 kV and 1 mA, and the charge neutralization system was used to obtain high-resolution spectra for the insulating materials such as polymers by reducing the surface charge. The spectra were taken with the electron emission angle at 0° to give a sampling depth of ∼100 Å. The sensitivity factors of individual elements were taken with the values from the standard Vision Library provided by the manufacturer, which were based on a combination of photoelectric cross-section, transmission function, and inelastic mean free path. Solid-State Nuclear Magnetic Resonance (NMR) Spectroscopy. Aromatic polyamide polymers for the solid-state NMR experiments were synthesized by interfacial polymerization, not on the polysulfone support, but in the beaker, according to the same recipes and procedure for the thinfilm formation of the individual TFC membranes. The interfacial reaction proceeded without stirring, and the fragments of the polymer films formed at the solution interface were retrieved, dried at 95 °C for 210 s, and precipitated in sodium carbonate solution (0.2 wt %). The solid polymers were suction-filtered off, washed extensively with freshwater and methanol to remove the unreacted monomer remnant and occluded salt, and dried at 50 °C in a vacuum oven. The NMR experiments were performed with a Bruker MSL-200 spectrometer (4.7 T and 200.13 MHz for 1H). The proton spin-lattice relaxation time in the rotating frame, T1F, was measured at temperatures from 200 K to 360 K by analyzing the magnetization decay after 1H 90°(x)spin-lock-τ pulse sequence. The pulse sequence was employed with 1H 90° pulse width of 4.5 µs and the repetition time of 5-6 T1 for the net magnetization to be completely relaxed. Typically, 20-25 different τ values were used, and FIDs were integrated in order to characterize the individual decay curves. Various temperatures were established by regulating the heat current in steady dry-air flow above room temperature and in steady cold nitrogen gas flow from the dewar below room temperature with accuracy of (0.5 K.

Results and Discussion Table 2 contains the RO performance data of water flux and salt rejection for all the aromatic polyamide thin-filmcomposite (TFC) membranes. The TFC membranes show a large increase in water flux, up 5-fold compared to FT-30, on VOL. 35, NO. 21, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4335

TABLE 2. Transport Characteristics and Surface Characteristics of Aromatic Polyamide TFC Membranes RO performance

max. water salt peak-to-valley mean surface surface flux rejection distance, height, roughness, area, S sample (gfd) (%) Rp-v (µm) Zh (µm) Rav (Å) (µm2) FT-30 A1 A2 A3 A4

15.2 49.2 51.4 68.7 76.9

96.4 95.1 93.1 89.6 87.4

0.470 0.556 0.602 0.604

0.146 0.232 0.243 0.246

423 553 567 580

167 194 203 209

a All the results were obtained with 2000 ppm NaCl in deionized water and at the operating pressure of 225 psi and the temperature of 25 °C.

the addition of dimethyl sulfoxide (DMSO) during interfacial polymerization. The salt rejection is not sacrificed considerably and is suitable for the TFC membranes used in lowpressure applications such as tap water purification, where high flux rather than rigorous rejection is required. The salt rejection of the A4 TFC membrane was 87%, although the water flux is highest among the five TFC membranes, which is beyond the practical range of the reverse osmosis applications. Thus, the A4 membrane is not appropriate for our purpose and is precluded from the further consideration. Figure 1 represents AFM surface images of all the TFC membranes with a projection area of 10 µm × 10 µm, in

which the unique and characteristic ridge-and-valley structure of the aromatic polyamide membranes is clearly shown. The bar at the left side of each image indicates the vertical deviations in the membrane surface; the white region is the highest and the black is the lowest. The AFM permits measurements of the distance variations on the surface with a line traversing the image as shown in each figure. This results in vertical height profiles of the morphological properties of membrane surfaces. Figure 2 compares the vertical height profiles along the horizontal line of 10 µm for all membranes. They are shown to differ in their surface line roughness, judging from the vertical deviation with respect to the scale of the ordinate as well as the numbers of peaks and valleys along the abscissa. The A3 TFC membrane possesses a greater number of pixels in the higher range of the vertical height, compared to the remaining three TFC membranes. Further statistical analysis is also possible on the entire projection area, where the maximum peak-tovalley distance, Rp-v, mean height, z, average roughness, Rav, and surface area, S, are usually determined. Rp-v is the difference in height between the highest and the lowest points within the selected areas:

Rp-v ) zmax - zmin

zj is given by the average height of the height profiles within the selected area:

FIGURE 1. AFM surface images of FT-30 (a), A1 (b), A2 (c), and A3 (d) TFC membranes. 4336

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 21, 2001

(3)

FIGURE 3. Concept of surface parameters; height (a), spacing (b), angle (c), distance (d), and surface area (e). FIGURE 2. Vertical height profiles of the four TFC membranes taken from the line traversing the individual AFM images. N

zj )

zn

∑N

(4)

n)1

Rav is the mean roughness of the surface relative to the center plane, that is the imaginary flat floating at the mean height, and is given by the average deviation of the data points (N) referenced to the average value of the data within the selected area:

Rav )

N

|zn - zj|

n)1

N

∑

(5)

The surface area, S, within the projection area is the sum of doubling the area of the triangles which consist of three sides of distance, d1, d2, and d3, and the distances between data points are calculated by measuring the height, h, spacing, s, and angle, θ. The resulting rectangular areas corresponding to individual data points are added (Figure 3). In this study, the quantitative analysis of the surface morphology was performed with three to five replication images for the individual membranes, and their arithmetic means are also summarized in Table 2. As shown in Table 2, there is an obvious correlation of the water permeability with the surface roughness and surface area of the TFC membranes. That is, the permeation flux increases in proportion to the increasing surface roughness and the enlarging surface area without a considerable loss of salt rejection. This trend of changes in surface characteristics with respect to the concentration of the substance added during interfacial polymerization is explained by considering the effect and the role of dimethyl sulfoxide (DMSO). DMSO has a solubility parameter of 12 (cal/cm3)1/2, whose value is between those of water (23.4) and n-hexane (7.3). Thus, DMSO in the aqueous solution phase works to increase miscibility of the aqueous and the organic phases by reducing the solubility difference of the two immiscible solutions and facilitates diffusion of diamine to the organic phase by regulating the interfacial tension between the two immiscible phases during interfacial reaction. This gives rise to fluctuation in the interface and, hence,

FIGURE 4. Proposed model for representing the effect of DMSO on the formation of surface morphology without (a) and with addition of additive (b). results in increasing surface roughness and enlarging surface area to have contact with water molecules, as depicted in Figure 4. The rougher surface and the larger surface area of the TFC membranes using DMSO during interfacial reaction in turn make it possible to have contact with more water molecules in the given projected area, which attributes to the higher permeability. In addition to the morphological structure, another important effect on the high permeation is the variation of the molecular structure and corresponding local motion in the TFC membranes with respect to the addition of DMSO on interfacial reaction. Recognizing the high density of amide linkage has been shown to be effective for high water permeability (9). The change in the chemical composition of the four TFC membranes was analyzed by X-ray photoelectron spectroscopy (XPS), especially with regard to the amide linkage with respect to the addition of DMSO. The XPS spectra of the four TFC membranes resulted in typical C 1s (285 ∼ 289 eV), N 1s (397.9 eV), and O 1s (531.6 eV) core levels for aromatic rings, amide groups, and carboxylic acids. Since the spectra are similar to those in the literature (18), they are not shown here. It is noted that the C 1s peaks of VOL. 35, NO. 21, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4337

TABLE 3. XPS Analysis of Atomic Concentrations and Relative Ratios of Elements for Aromatic Polyamide TFC Membranes sample

carbon atom %

oxygen atom %

nitrogen atom %

N/O

O/C

N/C

FT-30 A1 A2 A3

73.7 73.6 73.7 73.4

15.4 15.1 14.9 14.8

10.9 11.3 11.4 11.8

0.71 0.75 0.77 0.79

0.209 0.206 0.202 0.202

0.148 0.154 0.155 0.160

FIGURE 5. Chemical structure of repeat unit of aromatic polyamide.

TABLE 4. Composition Ratio of Repeat Units for Aromatic Polyamides sample

cross-linked portion with amide linkage (%)

linear portion with pendant COOH (%)

FT-30 A1 A2 A3

49 56 61 65

51 44 39 35

carbonyl carbon of carboxylic acid (known to exist at 288.7 eV) and of amide groups (288.2 eV) are not resolved and are overlapped as a shoulder on the aromatic ring carbon peak at 285.0 eV. Thus, it is rather difficult to calculate the variation of chemical composition (amide linkage) directly from the C 1s peaks. Instead, the relative concentrations of elements determined by XPS are fairly reliable and can be used for the calculation of the chemical changes in conjunction with DMSO. The relative concentrations of carbon, nitrogen, and oxygen are calculated with the following equation, and the results for the four TFC membranes are listed in Table 3

Ci )

Ai/Si

(6)

m

∑A /S j

j

j

where Ai is the photoelectron peak area of the element i, Si is the sensitivity factor for the element i, and m is the number of elements in the sample. Table 3 also shows the relative ratios of nitrogen to oxygen (N/O), oxygen to carbon, and nitrogen to carbon for the four TFC membranes. Following the chemical formula shown in Figure 5, the fractions of the cross-linked portion (m) possessing one more amide linkages and the linear portion (n) containing free pendant carboxylic acid groups can be calculated from the relative ratios of atomic concentration. In the calculation, m + n ) 1 and N/O ) (3m + 2n)/(3m + 4n) from chemical formula (C18H12N3O3) of the cross-linked portion and that (C15H10N2O4) of the linear portion. This calculation delivers the estimate of values of m and n fraction, respectively, as represented in Table 4. For the FT-30 TFC membrane, the ratio of the cross-linked amide to linear carboxylic acid formation is approximately 1:1, which is in good agreement with the previous study by others (19). Reid and Breton explain the water permeation in RO 4338

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 21, 2001

membranes by a hydrogen bond formation with functional groups of membrane surface and a formation of bound water (20). The bound water passes through membrane from one hydrogen-bonding site to another site in the membrane matrix under a pressure gradient, and the passage of a solute component that does not form a hydrogen bond is resisted. Interaction energy and stability of hydrogen-bonding formation between water and membrane surface would have an important effect on water permeability and salt rejection. Water can form hydrogen bonds with hydrophilic carboxylic acid groups or amide linkages. Both carboxylic acid group and amide functionality yield a shorter and stronger H-bond with water. According to experimental and theoretical results, the doubly bonded oxygen acts as a much more effective proton acceptor than does hydroxyl oxygen of carboxylic acid and NH group of amide linkage. Moreover, carboxylic acid groups form a slightly more stable and strong H-bond with water than hydrophilic amide groups (21). Higher stability of H-bond of carboxylic acid group with water would promise a more selective permeation but act as a resistance of the passage of the bound water from one site to another. In our cases, the addition of DMSO increases the number of hydrophilic amide linkage in the resulting TFC membranes, which enhance the water permeability without a considerable loss of salt rejection. On the basis of calculations from O/C and N/C ratios, a similar trend in the change of m and n values also results although the absolute values are somewhat different. Figure 6 depicts the variation of 1H solid-state nuclear magnetic resonance (NMR) magnetization, M(τ), as a function of delay time, τ, in a spin-lock pulse sequence for the four aromatic polyamides forming thin films at 300 K. The magnetization decays exponentially and are fitted by a nonlinear least-squares fit based on the following exponential function:

M(τ) ) M0exp(-τ/T1F)

(7)

The decays were fitted to within 95% confidence of the total signal by a single-exponential function, from which the time constant T1F, i.e., spin-lattice relaxation time in the rotating frame, was derived. The temperature dependence of the resulting T1F’s in the range of 200 K-360 K for all the aromatic polyamides is shown in Figure 7. The T1F’s versus reciprocal temperature exhibit a minimum, which indicates the most efficient relaxation; the T1F’s on the left side of the minimum indicate the faster motional states than at the minimum and vice versa. Thus, even the same value of T1F’s may be in very different motional states, depending on whether they locate on the left side (fast motion) or the right side (slow motion) of the minimum. To avoid such confusion and to relate T1F’s to molecular mobility, it is necessary to determine the correlation time, τc, on the basis of Bloembergen-Purcell-Pound (BPP) theory (22)

[

]

2.5τc τc 1.5τc 1 3 γ4p2 ) + + T1F 10 r6 1 + ω 2τ 2 1 + 4ω 2τ 2 1 + 4ω 2τ 2 0 c 0 c 1 c (8) where γ is the proton magnetogyric ratio, p is Planck’s constant divided by 2π, r is the distance between coupled spins, ωo is the Larmor frequency, and ω1 is the spin-lock field frequency, respectively. With eq 8, τc values are extracted by the nonlinear curve fitting of T1F data with respect to the minimum of T1F versus 1000/T plots and are shown in Figure 8. Recognizing the τc becomes longer as the local motion gets slower, the local mobility of the aromatic polyamides forming the TFC membranes increases in order from A3, A2, A1 to FT-30. This ordering is ascribed to the increase in the degree of cross-linking, and, hence, the number of amide

FIGURE 6. Magnetization intensity versus delay time at 300 K for the aromatic polyamides forming FT-30 (a), A1 (b), A2 (c), and A3 (d) TFC membranes.

FIGURE 7. Rotating-frame spin-lattice relaxation time, T1G, versus inverse temperature for the four aromatic polyamides.

FIGURE 8. Correlation time, τc, versus inverse temperature for the four aromatic polyamides.

linkages as the higher concentration of DMSO is used for the formation of thin films of the membranes. Thus, the NMR analyses agree well with and confirm the XPS results, in which there is an obvious correlation between molecular structure, local motion, and transport performance with reference to the RO characteristics in Table 2.

Acknowledgments The authors are grateful to the Korea Science and Engineering Foundation (KOSEF) for the support of this study through Hyperstructured Organic Materials Research Center (HOMRC). VOL. 35, NO. 21, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4339

Literature Cited (1) Cadotte, J. E.; Petersen, R. J.; Larson, R. E.; Erickson, E. E. J. Desalination 1981, 32, 25. (2) Cadotte, J. E. U.S. Patent 4,277, 344 (July 7, 1981). (3) Williams, M. E.; Bhattacharya, D.; Ray, R. J.; McCray, S. B. Membrane Handbook; Winston Ho, W. S., Sirkar, K. K., Eds.; Van Nonstrand Reinhold: New York, 1992. (4) Matsuura, T.; Sourirajan, S. Proc. Symp. on Advances in Reverse Osmosis and Ultrafiltration at 3rd Chem. Congr. North American Continent: Toronto, 1989. (5) Caetano, A.; Pinho, M. N. D.; Drioli, E.; Muntau, H. Membrane Technology: Applications to Industrial Wastewater Treatment; Dordrecht: Kluwer Academic Pub.: Boston, 1995. (6) Petersen, R. J. J Membr. Sci. 1993, 83, 81. (7) Chan, W. H.; Lam-Leung, S. Y.; Ng, C. F. Polym. Commun. 1991, 32, 501. (8) Hurndall, M. J.; Sanderson, R. D.; Jacobs, E. P.; VanReenen, A. J. Desalination 1993, 90, 41. (9) Kurihara, M. J. M. S. Pure Appl. Chem. 1994, A31, 1791. (10) Mukherjee, D.; Kulkarni, A.; Gill, W. N. J. Membr. Sci. 1994, 97, 231. (11) Kulkarni, A.; Mukherjee, D.; Gill, W. N. J. Appl. Polym. Sci. 1996, 60, 483.

4340

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 21, 2001

(12) Kulkarni, A.; Mukherjee, D.; Gill, W. N. J. Membr. Sci. 1996, 114, 39. (13) Kwak S.-Y.; Ihm, D. W. J. Membr. Sci. 1999, 158, 143. (14) Yamaguchi, T.; Ikeda, K. U.S. Patent 5,160, 619 (Nov. 3, 1992). (15) Hirose, M.; Ikeda, K. U.S. Patent 5,576, 057 (Nov. 19, 1996). (16) Hirose, M.; Ito, H.; Maeda, M.; Tanaka, K. U.S. Patent 5,614, 099 (Mar. 25, 1997). (17) Tomaschke, J. E. U.S. Patent 4,872, 984 (Oct. 10, 1989). (18) Clark, D. T.; Peeling, J.; Colling, L. Biochim. Biophys. Acta 1976, 453, 533. (19) Cadotte, J. E. In Materials Science of Synthetic Membranes; Lloyd, D. R., Ed.; American Chemical Society: Washington, DC, 1985. (20) Reid, C. E.; Breton, E. J., Jr. J. Appl. Polym. Sci. 1959, 1, 133. (21) Scheiner, S. Hydrogen Bonding: A Theoretical Perspective; Oxford University Press: Oxford, New York, 1997. (22) Bloembergen, N.; Purcell, E. M.; Pound, R. U. Phys. Rev. 1948, 73, 679.

Received for review February 12, 2001. Revised manuscript received July 12, 2001. Accepted July 23, 2001. ES010630G