Effect of Hydration of Polyamide Membranes on the Surface
Electrokinetic Parameters: Surface Characterization
by X-Ray Photoelectronic Spectroscopy
and Atomic Force Microscopy
M. J. Ariza,∗J. Benavente,∗E. Rodr´ıguez-Castell´on,†,1and L. Palacio‡
∗Departamento de F´ısica Aplicada and†Departamento de Qu´ımica Inorg´anica, Facultad de Ciencias, Universidad de M´alaga, 29071 M´alaga, Spain; and‡Departamento de Termodin´amica y F´ısica Aplicada, Facultad de Ciencias, Universidad de Valladolid, 47071 Valladolid, Spain
Received May 29, 2001; accepted October 29, 2001
The surface and the solid/liquid interface of two polyamide mem-branes, one experimental (B0) and one commercial (NF45), have been characterized by X-ray photoelectronic spectroscopy (XPS), atomic force microscopy (AFM), andζpotential, respectively. The surface roughness, determined by AFM data analysis, is different for the two membranes, and results show that the commercial NF45 membrane presents a much lower roughness than the experimen-tal B0 membrane. XPS data indicate that the surface of membrane NF45 is similar to that of pure polyamide, while membrane B0 contains a considerable amount of impurities. The homogeneity in depth of both membranes was also studied by determining the composition profile at different analysis angles. Streaming potential along the membrane surface or tangential streaming potential (TSP) measurements with NaCl solutions at different concentrations were carried out with both membranes to determine theζpotential and the electrokinetic surface charge density, and a correlation between membrane surface and interface parameters is made. Some differ-ences in atomic concentrations of membrane surface elements and X-ray photoelectronic spectra of the samples used in TSP measure-ments and after a drying process at 90◦C for 24 h can be observed when they are compared with those for fresh membranes. Elec-trokinetic parameters for membrane NF45 (TSP,ζ potential, and surface electrokinetic charge density) obtained from three different series of measurements strongly decrease as a result of membrane use, but for membrane B0 they are practically independent of the number of measurements. This difference in the electrokinetic be-havior of the two membranes has been related to the hydration process of the surface for each sample studied by XPS and AFM.
C
2002 Elsevier Science (USA)
Key Words:X-ray photoelectronic spectroscopy (XPS); atomic force microscopy; streaming potential; membrane hydration; depth profile.
INTRODUCTION
Membrane separation techniques are now commonly used in many industrial processes, and different materials and
mem-1To whom correspondence should be addressed.
brane structures are being investigated to obtain, for a given process, the highest retention and flux, retention and flux being two of the major parameters that characterize membrane perfor-mance (1, 2). These two parameters depend to a large extent on membrane morphology and surface chemistry, although in the case of transport of electrolyte solutions through membranes, the electrical membrane/solution interaction is also a relevant factor and, in turn, the surface charge of a membrane is another important parameter (3–8).
Membrane surface charge density can be determined from electrokinetic measurements (streaming potential and electroos-mosis) which allow determination of the zeta potential (ζ) or electrical potential at the shear plane. Moreover, other charac-teristic parameters for the solid phase (the membrane) such as the Gibbs free energy and the density of accessible sites per unit of membrane area can also be determined from electrokinetic parameters if an adsorption model is considered (9, 10).ζ po-tential is the parameter commonly used to study changes in the membrane surface due to the adsorption/deposition of particles (membrane fouling) or even chemical modifications able to af-fect the surface charged groups (11–13). However, significant differences in electrokinetic parameters have also been reported for similar membranes and hydrodynamic and chemical condi-tions even when no fouling or chemical changes exist, and these have been attributed to possible differences in measuring cells and procedures, membrane preservation, surface impurities, etc. (14, 15).
To study possible morphological and chemical changes in the membranes due to their use in contact with aqueous so-lutions at different salt concentrations, different surface analy-sis techniques can be used. Comprehensive characterization of membrane surfaces requires a multi-instrumental approach, as no individual surface analysis technique can ascertain both the chemical and morphological natures of their surfaces. Mem-brane surface characterization (topographic and chemical) can be carried out by scanning electron microscopy (SEM) and/or atomic force microscopy (AFM), when a higher spatial reso-lution is necessary, as can be conclude from the wide use of
149 0021-9797/02 $35.00
C
2002 Elsevier Science (USA) All rights reserved.
the latter technique in membrane characterization (16–18), and X-ray photoelectron spectroscopy (XPS) (19–21). Recently, the hydration of a commercial ion-exchange membrane, Nafion, was studied with two complementary techniques: AFM and small-angle X-ray scattering; both techniques indicated that the num-ber density of ionic clusters changes as a function of water con-tent (22, 23). On the other hand, the combined use of XPS and AFM allows the characterization of porous Nylon membranes coated with a crosslinked polysiloxane and modified by expo-sure to ultraviolet light (24); both techniques were also applied to study purple membranes deposited on gold (25).
The aim of this work is the surface characterization by different techniques of two composite polyamide/polysulfone membranes, one commercial and the other experimental. The arrangement of charge on the membrane surface and other char-acteristic membrane/solution interfacial parameters were inves-tigated by streaming potential measurements along the mem-brane surface or tangential streaming potential (TSP), while the surface chemical and topographic characterizations of the mem-branes were carried out by XPS, SEM, and AFM, respectively. Three series of TSP measurements were made with the mem-branes in contact with NaCl solutions at different concentra-tions (5×10−4≤C(M)≤10−2 M), keeping the membranes in distilled water for 48 h between two series of measurements. Changes in the electrokinetic parameters among the three series of measurements can be correlated with morphological differ-ences between both membranes and with the chemical changes observed in the membrane surfaces after being used in streaming potential measurements (polymer hydration).
MATERIALS AND METHODS
Materials
Two different types of membranes were studied: (i) a com-mercial composite polyamide/polysulfone nanofiltration mem-brane from FilmTec, called NF45; and (ii) an experimental composite polyamide/polysulfone membrane prepared at the Department of Analytical Chemistry, Universidad Aut´onoma de Barcelona (Bellaterra, Spain), and kindly supplied by Dr. M. Valiente and Dr. M. Mu˜noz, which will hereafter be called B0. The polysulfone membrane was obtained by using the phase inversion technique and the polyamide top layer was prepared by interfacial polymerization from an aqueous solution of 1,3-phenylenediamine. Finally, the composite membrane was water washed and dried at 60◦C for 10 min. A more detailed descrip-tion of membrane B0 is reported elsewhere (26).
Surface Characterization of Membranes
To examine the morphology of the membrane surfaces, a num-ber of scanning electron micrographs were recorded using a JEOL JSM-6400 scanning electron microscope. Flat samples were coated with a thin layer of gold before examination by mi-croscopy. A NanoScope Multimode Illa (Digital Instruments)
atomic force microscope was also used to study the surface of the samples. The membrane surfaces were imaged in a noncontact mode in air. Images were recorded and treated using NanoScope V4.23 software, which also provides roughness values (27). Sev-eral measurements (more than three) of each sample were carried out. The average roughness (Ra) and the standard deviation of the highness (Rq) values for the two membranes were obtained from the analysis of several pictures to obtain a mean value, taking into account that topographical properties can be slightly different in different areas of the membrane. The scanned areas measured 15×15µm2, a large enough membrane surface to obtain a constant value for the roughness without any depen-dence on the scanned areas (28, 29).
Surface chemical characterization was carried out by XPS. X-Ray photoelectron spectra were obtained using a Physical Electronics PHI 5700 spectrometer with nonmonochromatic MgKαradiation (300 W, 15 kV, 1253.6 eV) as excitation source. High-resolution spectra were recorded at different takeoff angles relative to the membrane surface, by a concentric hemispherical energy electron analyzer operating in the constant pass energy mode at 29.35 eV, using a 720-µm-diameter analysis area. Un-der these conditions the Au 4 f7/2line was recorded with 1.16 eV FWHM at a binding energy of 84.0 eV. The spectrometer energy scale was calibrated using Cu 2 p3/2, Ag 3d5/2, and Au 4 f7/2 photoelectron lines at 932.7, 368.3, and 84.0 eV, respectively. Charge referencing was done against aromatic carbon (C 1s 285.0 eV) (30). Membranes were mounted on a sample holder without adhesive tape and kept overnight at high vacuum in the preparation chamber before they were transferred to the analy-sis chamber of the spectrometer for their analyanaly-sis. Each spectral region was scanned several sweeps until a good signal-to-noise ratio was observed, and survey spectra in the range 0–1200 eV were also recorded at 187.85 eV of pass energy. For each mem-brane, XPS analysis was carried out three times using different samples. Membranes were irradiated separately and for a max-imum of 20 min to minimize X-ray-induced sample damage. The pressure in the analysis chamber was maintained lower than 5×10−6 Pa. The PHI ACCESS ESCA-V6.0 F software pack-age was used for acquisition and data analysis. A Shirley-type background was subtracted from the signals. Recorded spectra were always fitted using Gauss–Lorentz curves and following the methodology described in detail elsewhere (15, 20), to deter-mine more accurately the binding energy (BE) of the different element core levels. The spectra of C 1s, O 1s, and N 1s were fitted to two or more peaks to obtain a good fit of the spectra. Atomic concentration percentages of characteristic membrane surfaces elements were determined taking into account the cor-responding area sensitivity factor (31) for the different spec-tral regions measured. For scaling reasons, the intensity of each spectrum was normalized at its maximum value in all spectral figures.
An in-depth analysis was carried out by varying the takeoff angle (α) (30, 32, 33) taking into account the relation between the escape depth (d) and the photoelectrons mean free path (λ):
d ≤3λsinα. Although λ depends on electron kinetic energy (30), for simplicity we consider the value for the mean free path of the C 1s photoelectrons excited with MgKα(KE∼969 eV) in polyamide to be 3.2 nm (33, 34), which enables us to obtain depths between 2.5 and 9.3 nm by varying the analysis angle between 15◦and 75◦.
Electrokinetic Characterization of the Membrane/Solution Interface
Tangential streaming potential measurements were carried out with an EKA analyzer (Anton Paar, Graz, Austria), which basi-cally consists of: (i) a rectangular cell with a channel of well-defined dimensions (74×10-mm2 section and 300±10-µm thickness) created by the use of Teflon spacers between two membrane samples placed facing each other; (ii) a mechanical drive unit to produce and measure the pressure that drives the electrolyte solution from a reservoir into the measuring cell; and (iii) two Ag/AgCl electrodes, one at each end of the channel, to measure the induced streaming potential. Pressure was varied between 5×103and 3×104Pa. Electrokinetic measurements were carried out with the membranes in contact with NaCl so-lutions at different concentrations (5×10−4≤C(M)≤10−2), at room temperature (t =25.0±0.3◦C), and at standard pH (5.8±0.3). The streaming potential coefficient (φ=V/P)
was determined from the average of at least 10 measurements. To use the Fairbother–Mastin correction for surface conductivity (35), cell electrical resistance with the different NaCl solutions (Rc) and with a concentration high enough to assume no surface conductivity contribution (Ra, when C=0.1 M NaCl) were also measured using an ac bridge at 2.5-kHz constant frequency.
Three series of TSP measurements were carried out with each membrane for concentrations in the range previously indicated. After a complete series, the membrane was rinsed with running water until recovering distilled water conductivity; after that, before used in a new series of measurements, it was maintained for at least 48 h in distilled water.
RESULTS AND DISCUSSION
Surface Characterization
The scanning electron micrographs of membranes NF45 and B0 (Fig. 1) show the presence of different structures and rough-ness on the polyamide surface. From AFM images it is possible to determine the surface roughness in addition to other param-eters such as size, shape, and density of the pores (27, 36–37). Figure 2 shows two-dimensional images of membranes NF45 and B0. As well as their SEM images, the grey color scale indi-cates the highness of the sample. It can be observed that there is not a well-defined pore structure in both membranes, but their surfaces are more or less rough due to the polymerization pro-cess. The structures observed by AFM are similar to those ob-served by SEM, but the former technique allows determination of the surface roughness since AFM images are in three dimen-sions. There are several definitions of surface roughness using
FIG. 1. Scanning electron micrograph of membranes NF45 and B0.
the values of the tip highness in each point of the image (Zi)
over a reference baseline (Z ); the most common one used is the average roughness (Ra), which is defined as (27)
Ra= 1
n
n
i=0
|Zi−Zm|, [1]
where Zmis the mean value of Zi. The standard deviation of the
highness (Rq) is obtained from
Rq=
(Zi−Zm)2
FIG. 2. AFM bidimensional images of membranes NF45 and B0.
The Raand Rq values are 15±2 and 21±2 nm for mem-brane NF45 and 119±5 and 153±6 nm for membrane B0. Therefore, the surface of the commercial NF45 membrane has a much lower roughness than the surface of the experimental B0 membrane. Three-dimensional AFM images can provide sim-ilar information (Fig. 3), but now it is possible to see that the shapes of the roughness crest in the two membranes are slightly different, which must be due to differences in the process of manufacturing the commercial and experimental membranes. It should be emphasized that the different surface roughnesses observed can affect the behavior of the membrane in contact with electrolyte solutions, particularly the membrane electrical parameters obtained by tangential streaming potential
measure-FIG. 3. Tridimensional AFM images of membranes NF45 and B0.
ments, as higher roughness represents more surface in contact with the electrolyte solutions. In fact, the effect of high surface roughness on colloidal fouling of membranes has already been reported in the literature (38).
The composite membranes studied consist of a thin layer of polyamide (≤1µm) deposited on a polysulfone porous support (26, 39). So, the chemical information obtained through XPS analysis is derived only from the near surface of the top layer, and the polysulfone layer is not probed; thus, the main chemical elements studied by XPS were C, N, and O (the polyamide char-acteristic elements). Table 1 shows their atomic concentration percentages in both membranes at a 45◦ takeoff angle (NF45 and B0 samples). Membrane NF45 also contains S, but with
TABLE 1
Surface Composition by XPS Analysis of Membranes NF45 and B0 Fresh and Used in Electrokinetic Measurements
Membrane % C % O % N O/C N/C N/O
NF45 74.4±0.8 13.7±0.5 11.9±0.3 0.185±0.009 0.159±0.006 0.87±0.05 NF45u 69.5±0.2 20.5±0.1 10.0±0.2 0.295±0.002 0.144±0.003 0.49±0.01 NF45u-(d) 69.4±0.5 20.4±0.5 10.2±0.1 0.294±0.009 0.147±0.003 0.50±0.02 B0 77.6±0.8 9.8±0.4 12.6±0.4 0.126±0.006 0.162±0.006 1.28±0.09 B0u 71.0±1.0 18.9±0.9 10.0±0.5 0.266±0.016 0.141±0.009 0.53±0.05 B0u-(d) 70.1±0.3 19.7±0.9 10.2±0.3 0.281±0.014 0.145±0.005 0.52±0.04
an atomic concentration lower than 0.5%, and there is a slight excess of oxygen with respect to the corresponding theoretical ratio for polyamide (N/O=1), which indicates a low content of impurities on the surface of this membrane. The surface of membrane B0 contains S, P (<0.5%), and Cl (<1%) in addition to the elements characteristic of the polyamide. This membrane exhibits a deficiency of oxygen and an excess of nitrogen with respect to the theoretical value of the polyamide. The excess of nitrogen and the presence of phosphorus and chlorine can indicate that solvents and washing agents were not completely eliminated during the process of manufacturing the experimental B0 membrane. Table 2 lists the different bands (four for carbon, two for oxygen, and three for nitrogen) obtained after curve fit-ting. The selection of these bands was done taking into account references that include XPS information on carbon-, oxygen-, and nitrogen- containing polymers (40–50). Since polyamide is the only source of nitrogen for the membranes, it can be dis-tinguished from amide groups of neutral polymers (N2), pro-tonated or hydrogen-bonded amide groups (N3), and terminal amine groups (N1). The multiregion spectra of C 1s, O 1s, and N 1s for both membranes are shown in Fig. 4, and Table 3 includes a summary of the XPS curve fitting parameters.
Taking into account the previous values, we can assert that the surface of membrane NF45 is similar to that of a pure polyamide. By using the percentage area corresponding to the O1 and N2 peaks a N/O molar ratio of 0.928 was obtained. This value is closer to the theoretical value of 1 than that obtained by using the complete area of the O 1s and N 1s signals. The percentage of carbon corresponding to the C4 peak is also practically co-incident with the amounts of amide oxygen and nitrogen. This
TABLE 2
C 1s, O 1s, and N 1sPeaks Used in the Curve Fitting of the Mutiregion Signals for Membranes NF45 and B0
Band BE (eV) Type of atom Reference BE (eV)
C1 285.0 Aromatic/aliphatic 285.0 (40–42)
C2 285.7±0.1 β(–C–COOR/H, –C–CONR/H) 285.7 (40, 41, 43)
C3 286.7±0.1 –C–O, –C–N, –C–O(H), HC–(C)3 286.1–287.0 (40, 42–44)
C4 288.4±0.2 –C==N, O==C–N, O–C–O 287.1–288.1 (40, 43, 45)
O1 532.0±0.3 C==O, O==C–N, C–O(H) 532.2–532.8 (29, 36, 44, 45)
O2 533.4±0.2 H--O==C–N, O==C–O, 533.6–534.3 (20, 31, 43)
N1 399.2±0.1 –C==N–, R-NH2 398.5–398.8 (44, 48, 49)
N2 400.4±0.1 O==C–N, C==NH 399.8–400.2 (45, 49)
N3 401.9±0.2 O==C–N--H, R-N+H3, R-N+H2R 401.0–401.9 (31, 49, 50)
means that 11% of the material studied by XPS corresponds to amide groups. The C2 peak is assigned toβcarbons of the acidic groups used in the interfacial polymerization. The molar ratio C2/C4 is very close to 0.5, indicating that the acid monomer used in the interfacial polymerization contains three acid groups.
As was previously indicated, membrane B0 presents a con-siderable amount of impurities. The amount of amide groups calculated from the C4 and O1 peaks is about 7%, whereas if the N2 peak is used, a higher value, 11%, is obtained. Since the adsorbed impurities can contain the three chemical elements of polyamide, it is difficult to give a value for the proportion of amide groups on this membrane.
To know the homogeneity in depth of the membrane sur-face, several spectra were taken at different analysis angles. The variation of atomic concentration as a function of the probed depth is shown in Fig. 5. The shapes of the peaks at different analysis angles were also studied by the corresponding curve fitting analysis. The composition profile of membrane NF45 in-dicates the membrane is homogeneous along the depth studied, except in the more superficial region (<3 nm) where a defi-ciency of nitrogen and an excess of carbon were observed. By curve fitting, the percentage of carbon obtained from C3 and C4 peaks (basically C–O and amide groups, respectively) decreases, and that obtained from the C1 peak (aliphatic and aromatic car-bons) increases. These results indicate the presence of aliphatic carbon and oxygen-rich groups on this membrane surface due to contamination. On the other hand, the N1 peak becomes slightly more intense at low analysis angle, which is indicative of a higher density of terminal chain groups on the surface of this membrane. The percentages of amidic carbon, oxygen, and
FIG. 4. C 1s, O 1s, and N 1s core level spectra for NF45 and B0 membranes.
nitrogen atoms, obtained from the C4, O1, and N2 peaks areas, are very similar at all the angles studied (∼11%) except at 15◦ (∼8%).
The profile of membrane B0 also seems to be very homo-geneous (Fig. 5). The shapes of the peaks at different angles are very similar, except at low angle (15◦), where C3 and C4 peaks lose intensity and the intensity of the O3 peak slightly increases in comparison to that of the O2 peak. These results suggest the presence of aliphatic carbons, water molecules, and other species derived from atmospheric contamination on the top surface of this membrane.
Correlation between the membrane surface and interfacial electrokinetic parameters, which are obtained with the mem-branes in contact with aqueous solutions, is one of the objectives of this article, as is discussed in the next section. Then
possi-TABLE 3
Curve Fitting Summary of Different XPS Peaks for Membranes NF45 and B0
NF45 B0
Banda FWHM % Area FWHM % Area
C1 1.80 37.2 1.82 45.5
C2 1.71 31.5 1.76 39.0
C3 1.40 14.4 1.65 7.6
C4 1.70 16.9 1.57 7.9
O1 1.84 80.9 1.78 73.9
O2 1.90 19.1 2.00 26.1
N1 1.68 6.3 1.70 7.3
N2 1.68 86.5 1.84 80.8
N3 1.68 7.2 2.00 11.9
aBE and band descriptions are included in Table 2.
ble chemical modifications of membrane surfaces as a result of their use in electrokinetic measurements were monitored by XPS. Table 1 also shows the atomic concentration percentages obtained at a 45◦ takeoff angle for the samples used (NF45u and B0u), and those used and dried at 90◦C for 24 h (NF45u-(d) and B0u-(NF45u-(d)). In comparison with the fresh membranes, the used samples exhibit an increase in oxygen content (O/C≈0.3, N/O≈0.5); similar results were obtained for the heated sam-ples. In the NF45u membrane there are other elements, such as Na+and Cl−(which appear mainly at the analysis angle of 15◦); the atomic concentration of these elements was lower than 1%, and they are attributed to the contact of membrane with NaCl aqueous solutions. However, the surface of sample B0u
FIG. 5. Variation of the atomic concentrations of C, O, and N as a function of the depth probed for membranes NF45 and B0.
FIG. 6. C 1s, O 1s, and N 1s core level spectra for membranes NF45 (a) and B0 (b), used and dried in an open atmosphere for a week (NF45u, B0u), and used and dried at 90◦C for 24 h (NF45u(d), B0u(d)).
does not contain other elements found in membrane B0 (for in-stance, phosphorus), which indicates that the continuous use of the membrane with aqueous solutions has cleaned its surface of remaining compounds used in the manufacturing process. Moreover, the molar ratio N/C decreases probably due to the elimination of aliphatic carbons from contamination or to a re-arrangement of the polymeric chain in such a way that amide groups are located nearer the surface, leaving the aromatic ring inside.
Figure 6 shows the C 1s, O 1s, and N 1s core level spectra for the different samples studied. As can be observed, the shape of the spectra of the polyamide characteristic elements is mod-ified after contact with aqueous solutions. In the C 1s signal, a decrease in the aliphatic or aromatic carbon component (peak at lower BE) and an increase in the amidic carbon component were observed. This relative increase is more notorious in the case of the B0 membrane than in the NF45 membrane. Perhaps, this occurs because membrane B0 has more surface impurities than membrane NF45 as was previously indicated, and the cleaning effect enhanced the relative intensity of the amidic carbon peak. More interesting are the modifications observed in the O 1s signal in both membranes on contact with aqueous solutions. In both cases, the intensity of the peak at higher BE (O2) increases; this effect is so high in membrane B0 that the maximum of the O 1s signal shifts to higher BE due to the presence of adsorbed water (40). However, the shape of the O 1s signal in the used and dried (B0u-(d)) sample was not modified on heating at 90◦C, while in the NF45u-d sample the relative intensity of O2 slightly decreased due to the thermal treatment. These results show the
adsorption of water in both polyamide membranes during their use in streaming potential measurements, although in the exper-imental B0 membrane, which has a larger surface roughness, a larger amount of water is permanently retained in comparison with the NF45 membrane. For this reason, a portion of the ad-sorbed water can be more easily evolved on drying due to the lower surface roughness of membrane NF45. As expected, the N 1s signal of both membranes remains practically unchanged after the different modifications and can be used as a reference for the width and position of the different peaks.
Membrane/Solution Interface Characterization
Streaming potential measurements are commonly used for electrical characterization of solid/liquid interfaces. When an electrolyte solution is forced to move along to a charged solid wall by an external pressure (P), the electrical potential or
“streaming potential” (øst) between the two ends of the sys-tem can be measured (51). The hydraulic pressure causes move-ment of liquid and thus ions are stripped off along the shear plane, and a streaming current is formed; due to charge accu-mulation at the downstream side, an electrical field is generated that causes a backflow of ions until a steady state is reached. The measurable electrical potential difference between the two ends of the solid/liquid system,øst, gives direct information about the electrokinetic charge at the shear plane (52). Figure 7a shows the linear relationships between tangential streaming potential and applied pressure for the polyamide active layer of NF45 and B0 membranes measured at different salt concentrations. From
FIG. 7. (a) Streaming potential versus applied pressure for NF45 () and
B0 () membranes in different NaCl solutions. (b) Concentration dependence for streaming potential coefficient,φ=δV/P, for NF45 () and B0 () membranes.
the slopes of these straight lines the streaming potential coeffi-cient,φ=(øst/P), was obtained, and its variation with salt
concentration is shown in Fig. 7b. As can be seen, for both mem-branes|φ|values strongly decrease when the salt concentration increases.
A simple relation between the streaming potential and the potential at the shear plane or zeta potential (ζ) can be obtained by the Helmholtz–Smoluchowski equation (51),
ζ =(λ0η/ε0εr)(øst/P), [3]
whereλ0andηare the liquid conductivity and viscosity, respec-tively, ε0 is the permittivity of vacuum, andεr is the relative dielectric constant of the electrolyte solution. Equation [3] is valid only for weakly charged walls and wide channels or pores, when the surface conductivity and double layer overlapping can be neglected. In other cases, the surface conductivity (λs), which takes into account the contribution of the fixed charges to the shear plane, must be considered (51, 52). Althoughλscannot be determined experimentally, theζ potential can be obtained by using the Fairbother–Mastin approximation if the resistances at
the concentrations studied and at high concentrations (Rcand
Ra, respectively) are known (49),
ζ =([λ0+(λs/h)]η)/ε0εr)(øst/P)
=(η/ε0εr)
λa 0Ra
Rc
(øst/P), [4]
where h is the half-thickness of the channel,λa0is the conductiv-ity of the solution at high concentration, and the other parame-ters have already been defined. Variation ofζ potential with salt concentration is shown in Fig. 8a. As can be seen, membrane B0 presents a more negative character than membrane NF45 for the whole range of concentrations.
Surface charge density (σs) can be determined fromζpotential values by (51)
σs=(2RTε0εrk/z F ) sinh(Fζ/2RT ), [5]
where k is the Debye length, and R and F are the gas and Faraday constants, respectively. Figure 8b shows variation of electroki-netic surface charge density (σs) for the polyamide active layer of both membranes as a function of the NaCl concentration. It can be observed that σs values do not show any clear con-centration dependence, and their average values for the concen-tration range studied are listed in Table 4. The electronegative character of both membranes agrees with their weak cation-exchange behavior already obtained from membrane potential measurements (39, 53). The slight electrokinetic differences between both membrane samples (Table 4) can be associated
FIG. 8. Concentration dependence for (a) zeta potential and (b) electroki-netic surface charge density. () NF45, () B0.
TABLE 4
Average Values of the Electrokinetic Charge Density (σS) in C/m2
for NF45 and B0 Membranes Obtained from Different Series of Measurements
σS(C/m2)
Membrane First series Second series Third series NF45 −(1.8±0.5)×10−3 −(1.5±0.7)×10−3 −(0.71±0.08)×10−3
B0 −(2.8±0.4)×10−3 −(2.7±0.2)×10−3 −(2.6±0.2)×10−3
with the higher roughness and impurity content of membrane B0.
The effect of membrane hydration on electrokinetic param-eters was also studied. Figures 9 and 10 show a comparison of concentration dependence for both streaming potential coef-ficient and electrokinetic charge density for membranes NF45 and B0 determined from different series of measurements. Av-erageσsvalues for the whole range of concentrations for each series of measurements and membrane are also summarized in Table 4. As can be seen, electrokinetic parameters for mem-brane NF45 strongly decrease as a result of memmem-brane use, but for membrane B0 they are practically independent of the num-ber of measurements. The differences found in the behavior of the polyamide layer of both membranes could be explained by taking into account AFM and XPS results. Stabilization of the membrane/solution interface can be achieved just in the first series of measurements for membrane B0 due to its high sur-face water adsorption previously determined from XPS results; however, for membrane NF45, polyamide hydration must
oc-FIG. 9. Variation of the streaming potential coefficient–concentration curves for the different series of measurements for NF45 and B0 membranes.
FIG. 10. Variation of the electrokinetic charge density–concentration rela-tionship for the different series of measurements for NF45 and B0 membranes (symbols have the same meaning as in Fig. 9).
cur slowly, according to its lower roughness, because of which changes in the electrokinetic parameters would be taking place during measurements.
Differences in membrane hydration and surface roughness could explain the diversity in the values of electrokinetic param-eters reported in the literature for similar membranes studied un-der the same hydrodynamic conditions and solution chemistry. According to this, XPS and AFM are two techniques that can help elucidate this observed diversity.
CONCLUSIONS
Characterization of both the surface and the solid/liquid inter-face of two nanofiltration membranes, NF45 and B0, has been carried out to correlate differences in interfacial parameters and behavior with chemical and/or morphological differences of the membrane surfaces. XPS data do not show significant chemical differences between the membranes but an increase in the oxy-gen content of samples of both membranes after contact with aqueous solutions was observed. This fact is attributed to poly-mer hydration, and it was not affected by membrane heating.
Strong differences in membrane surface roughness were de-termined from AFM measurements. Results show that the rough-ness of the experimental B0 membrane is around eight times higher than that of the commercial NF45 membrane. Changes in the electrokinetic parameters for membrane NF45 after suc-cessive measurements are attributed the a slow hydration of this membrane when compared with B0, in agreement with its lower roughness.
ACKNOWLEDGMENT
We thank CICYT (Project MAT2000-1140) for financial support. REFERENCES
1. Mulder, M., “Basic Principles of Membrane Technology.” Kluwer Academic, Dordrech, 1992.
2. Cheryan M., “Ultrafiltration Handbook.” Technomic, PA, Penn., 1986. 3. Anderson, J. L., and Kuh, W. H., J. Colloid Interface Sci. 59, 149 (1977). 4. Demish, H. H., and Pusch, W., J. Colloid Interface Sci. 76, 2 (1980). 5. Benavente, J., and Fern´andez-Pineda, C., J. Membr. Sci. 23, 1221 (1985). 6. Kimura, Y., Lim, H. J., and Ijima, T., Angew. Makromol. Chim. 136, 151
(1986).
7. Nysstr¨om, M., Lindstr¨om, M., and Matthiosson, E., Colloid Surf. 436, 297 (1989).
8. Benavente, J., and Jonsson, G., Sep. Sci. Technol. 32, 1699 (1997). 9. Benavente, J., Hern´andez, A., and Jonsson, G., J. Membr. Sci. 80, 285
(1993).
10. Hern´andez, A., Mart´ınez, F., Mart´ın, A., and Pr´adanos, P., J. Colloid
Inter-face Sci. 173, 284 (1995).
11. Le Bolay, N., and Ricard, A., J. Colloid Interface Sci. 170, 154 (1995). 12. Elimelech, M., Zhu, X., Childress, A. E., and Hong, S., J. Membr. Sci. 127,
101 (1997).
13. Ulbricht, M., Richaou, K., and Kamusewitz, H., Colloid Surf. 138, 353 (1988).
14. Wilbert, M. C., Delagoh, S., and Pellegrino, J., J. Membr. Sci. 161, 247 (1999).
15. Ariza, M. J., Ca˜nas, A., and Benavente, J., Surf. Interface Anal. 30, 425 (2000).
16. Iwata, H, Hirata. I., and Ikada, Y., Macromolecules 31, 3671 (1998). 17. Bowen, W. R., Hilol, N., Lovitt, R. M., Sharif, A. O., and Willians, P. M.,
J. Membr. Sci. 126, 77 (1997).
18. Hern´andez, A., Calvo, J. J., Pr´adanos, P., Palacio, l., Rodr´ıguez, M. L., and de Soja, J. A., J. Membr. Sci. 137, 89 (1998).
19. Fontyn, M., Bijsterbosch, B. H., and van’t Riet, K., J. Membr. Sci. 36, 141 (1987).
20. Ariza, M. J., Rodr´ıguez-Castell´on, E., Rico, R., Benavente, J., Mu˜noz, M., and Oleinikova, M., J. Colloid Interface Sci. 226, 151 (2000).
21. Ariza, M. J., Rodr´ıguez-Castell´on, E., Rico, R., Benavente, J., Mu˜noz, M., and Oleinikova, M., Surf. Interface Anal. 30, 430 (2000).
22. James, P. J., Elliot, J. A., Mcmaster, T. J., Newton, J. M., Elliot, A. M. S., Hanna, S., and Miles, M. J., J. Mater. Sci. 35, 5111 (2000).
23. James, P. J., Mcmaster, T. J., Newton, J. M., and Miles, M. J., Polymer 41, 4223 (2000).
24. Ouyang, M., Muisener, R. J., Boulares, A., and Koberstein, J. T., J. Membr.
Sci. 177, 177 (2000).
25. Brizzolara, R. A., Boyd, J. L., and Tate, A. E., J. Vac. Sci. Technol. A 15, 773 (1997).
26. Oleinikova, M., Mu˜noz, M., Benavente, J., and Valiente, M., Langmuir 16, 716 (2000).
27. Palacio-Mart´ınez, L., “Caracterizaci´on Estructural y Superficial de Mem-brans Microporosas,” Ph.D. thesis, Universidad de Valladollid, Valladolid, 1998.
28. Schultze, R. D., Possart, W., Kamusewitz, and H., Bischof, C., J. Adhesion
Sci. Technol. 3, 39 (1989).
29. Palacio, L., Pr´adanos, P., Hern´andez, A., Ariza, M. J., Benavente, J., and Nystr¨om, M., Appl. Phys. A 73, 555 (2001).
30. Brigg, D., Seah, and M. P., “Practical Surface Analysis,” 2nd ed, Vol. 1: “Auger and X-Ray Photoelectron Spectroscopy,” Wiley, Chichester, 1995. 31. Moulder, J. F., Stickle, W. F., Sobol, P. E., and Bomben, K. D., in “Handbook of X-Ray Photoelectron Spectroscopy” (J. Chastain, Ed.), Perkin–Elmer, Minneapolis, 1992.
32. Lukas, J., Richau, K., Schwarz, H. H., and Paul, D., J. Membr. Sci. 106, 281 (1995).
33. Pjipers, A. P., and Meier, R. J., Chem. Soc. Rev., 28 (1999).
34. Lukas. J., and Jezek, B., Collect. Czech. Chem. Commun. 48, 2909 (1983). 35. Fairbrother, F., and Mastin, H., J. Chem. Soc. 125, 2319 (1924). 36. M¨ockel, D., Staude, E., and Guiver, M. D., J. Membr. Sci. 158, 63 (1999). 37. Stamatialis, D. F., Dias, C. R., and de-Pinho, M. N., J. Membr. Sci. 160,
235 (1999).
38. Elimelech, M., Zhu, X., Childress, A. E., and Hong, S., J. Membr. Sci. 127, 101 (1997).
39. Ariza, M. J., Ca˜nas, A., and Benavente, J., Surf Interface Anal. 30, 425 (2000).
40. L´opez, G. P., Castner, D. G., and Ratner, B. D., Surf. Interface Anal. 17, 267 (1991).
41. France, R. M., and Short, R. D., Langmuir 14, 4827 (1998).
42. Shakesheff, K. M., Evora, C., Soriano, I., and Langer, R., J. Colloid Interface
Sci. 185, 538 (1997).
43. Davies, M. C., Lynn, R. A. P., Hearn, J., Paul, A. J., Vickerman, J. C., and Watts, J. F., Langmuir 11, 4313 (1995).
44. Sniss, A., and Matar, S. F., J. Chem. Phys. 111, 9678 (1999). 45. Burrel, M. C., and Chera, J. J., Surf. Interface Anal. 27, 811 (1999). 46. Buchwalter, L. P., Silverman, B. D., and Witt, L., J. Vac. Sci. Technol. A 5,
226 (1987).
47. Clark, D. T., and Harrison, A., J. Polym. Sci. Polym. Chem. Ed. 19, 1945 (1981).
48. Snauwaert, P., Lazzaroni, R., Riga, J., and Verbist J. J., J. Chem. Phys. 92, 2187 (1990).
49. Schick, G. A., and Sun, Z., Langmuir 10, 3105 (1994). 50. Jansen, R. J. J., and Van-Bekkum, H., Carbon 33, 1021 (1995). 51. Hunter, R. I., “Zeta Potential in Colloid Science: Principles and
Applica-tions.” Academic Press, London, 1988.
52. Lyklema, J., “Fundamentals of Interface and Colloid Science.” Academic Press, London, 1993.
53. Ariza, M. J., and Benavente, J., J. Membr. Sci. 190, 119 (2001). 54. Benavente, J., and Ca˜nas, A., J. Membr. Sci. 156, 2411 (1999).
