Modification of active and porous sublayers of aged polyamide/polysulfone
composite membranes due to HNO
3
treatment: Effect of treatment time
J. Benavente
∗, M.I. Vázquez, R. de Lara
Grupo de Caracterización Electrocinética de Membranas e Interfases, Departamento de Física Aplicada I, Facultad de Ciencias, Universidad de Málaga, E-29071 Málaga, Spain
Received 25 July 2005; accepted 6 October 2005 Available online 17 November 2005
Abstract
Changes in electrical and transport parameters for aged composite polyamide/polysulfone membrane samples (PAC) and their porous support layers (PSU) as a result of chemical treatment (immersion in 1 M HNO3solution) at four different times (12 ht72 h) have been obtained. Salt permeability, ion transport number, and membrane electrical resistance for the treated samples were determined from salt diffusion, membrane potential, and impedance spectroscopy measurements, which were carried out with the membranes in contact with NaCl solutions at different concentrations and compared with those determined for fresh and aged nontreated samples. Results show the strong effect of aging on membrane parameters, particularly the decrease in salt permeability (Ps) and the increase in membrane electrical resistance (Rm), while ion transport number is hardly affected by aging, chemical treatment, or treatment time. Results show how the compaction of the porous structure causes by aging (dried membrane matrix structure) can be partially reduced by HNO3treatment, and they also allow the estimation of 24-h treatment as the optimum time (higher salt permeability and lower membrane electrical resistance), mainly for the polysulfone support layer. The use of equivalent circuits in the analysis of impedance spectroscopy data allows separate estimation of the electrical resistance associated with each sublayer of the composite PAC membrane samples. On the other hand, chemical changes in the active top layer of the PAC membrane (polyamide active layer) were obtained from XPS analysis, which show some modifications in the atomic concentration percentages of the polyamide characteristic elements as a result of acidic treatment time, which are more significant after 72-h acidic immersion.
2005 Elsevier Inc. All rights reserved.
Keywords:Composite and porous membranes; Chemical treatment; Diffusion; Impedance spectroscopy; Transport numbers; XPS
1. Introduction
Membrane systems are nowadays used in many separation processes where volume, solute, and ion fluxes are originated by gradients of pressure, concentration, and electrical poten-tial, respectively[1,2]. Composite and asymmetric membranes are used in desalting processes under pressure gradient such as nanofiltration and reverse osmosis due to their high salt rejec-tion, volume flux, and mechanical stability. These membranes can be considered as multilayer systems coupled in series by infinitely thin layers of aqueous solutions in local equilib-rium with both adjacent layers [3]. For simplicity, composite membranes are commonly considered as two-layer systems[4]
* Corresponding author. Fax: +34 952 13200.
E-mail address:[email protected](J. Benavente).
formed by a dense and thin active layer (less than 1 µm thick-ness) and a thick and porous support, usually an ultrafiltration membrane (150–200 µm thickness, including the nonwoven structure for membrane reinforcement). It is assumed that the active layer controls membrane selectivity and fluxes, although the porous support can also slightly affect them[4,5], but three-layer (or even multithree-layer) models that include an intermediate or transition sublayer connecting dense and porous layers can also be considered[6,7]. In the case of composite membranes, aging, fouling, and cleaning agents can differently affect each sublayer of a composite membrane, depending on the material and structure. Information about typical membrane materials, characterization methods, and changes in transport parameters due to membrane fouling and aging, preservation conditions, cleaning procedures, and/or chemical resistance can be found in the literature [8–11]. Particularly, modification in the
ac-0021-9797/$ – see front matter 2005 Elsevier Inc. All rights reserved.
tive and support layers of different samples of a composite polyamide/polysulfone reverse osmosis membrane as a result of aging and chemical treatments by immersing the membrane in 1 M HCl, HNO3, or NaOH solution were studied in a previous work[12]. A decrease of around 40% in water and salt perme-abilities for the aged sample was obtained which was related to the compaction of the polymer chains associated to drier mem-brane structure (loss of humidic substances), but it was partially reduced when the samples were treated with both acidic solu-tions. These results are in agreement with the hydrophilizing effect of inorganic acids reported by different authors in the case of polyethylene and aromatic polyamide membranes[13, 14], which indicate the following hydrophilizing efficiency se-quence: HNO3>H2SO4>H3PO4.
In this paper, a systematic study of changes caused in aged samples by chemical treatment (immersion in 1 M HNO3 so-lution) as well as the time of treatment (0, 12, 24, 48, and 72 h) for a composite polyamide/polysulfone membrane (PAC) and its porous polysulfone support (ultrafiltration PSU mem-brane) is carried out. Transport and electrical parameters (salt permeability, ion transport number, and membrane electrical re-sistance) were obtained from electrochemical measurements, which were performed with the membrane samples in con-tact with NaCl solutions at different concentrations. To corre-late changes in the electrical resistance of the whole compos-ite PAC membrane with those associated with each sublayer (dense and porous layers), impedance spectroscopy measure-ments were also made. Impedance spectroscopy (IS) is an ac technique for electrical characterization of material and inter-faces based on electrical impedance measurements carried out for a wide range of frequencies (10−6–109 Hz), and it is a successful tool for determining the electrical properties of het-erogeneous systems formed by a series array of layers with dif-ferent electrical/structural properties using the impedance plots and the equivalent circuits as models, where the different circuit elements are related to structural/transport properties of the sys-tems[15–17]. In the case of composite membrane/electrolyte systems (or membrane/electrolyte systems), the analysis of the impedance plots allows separate evaluation of the electrical contribution of each sublayer: electrolyte between the mem-brane surface and the electrodes, dense and porous sublayers
[17–19]. A comparison of the results obtained for fresh, aged, and treated samples shows the effect of age and treatment time on the membrane structure and the possibility of determin-ing an optimum time for HNO3treatment. On the other hand, X-ray photoelectron spectroscopy (XPS) characterization of the polyamide active layer of PAC samples was also considered in order to determine possible chemical modifications in the polyamide surface due to the time of immersion in the acidic solution.
2. Material and methods
2.1. Material
Two commercial membranes kindly submitted by PRIDESA (Barcelona, Spain) were studied. One of them was a
com-posite polyamide/polysulfone membrane for reverse osmosis (membrane PAC) and the other was a polysulfone ultrafiltra-tion membrane used as a porous support in PAC manufactur-ing (membrane PSU). Samples of both membranes taken from packets that had already been opened for 2 years (without any special preservation) were chemically treated by immersion in 1 M HNO3solution for different times: 0, 12, 24, 48, and 72 h. These membranes will hereafter be named PAC-X and PSU-X, respectively, where X represents the treatment time. In this con-text, the differences existing among PSU-X membranes and the porous sublayer of the PAC-X samples due to the pres-ence of the polyamide top layer in the latter (no direct contact exists between both sides of the polysulfone sublayer and the acid solution) should be pointed out. For comparison, to see the effect of age and treatment, results with fresh samples will also be presented (PAC-F and PSU-F membranes, respectively). Hydraulic permeability for fresh and aged samples is [12]
LPAC-F
p =7.2×10−12 m/s Pa, LpPAC-0=3.8×10−12 m/s Pa,
LPSU-F
p =11.2×10−12m/s Pa,LpPSU-0=1.4×10−12 m/s Pa. According to these values, higher effect of aging on ultrafil-tration membranes than on composite membranes should be expected.
Electrochemical measurements were carried out with aque-ous NaCl solutions at different concentrations (10−3M< c <
5×10−2M), at room temperaturet=(25.0±0.5)◦C and stan-dard pH (6.0±0.2). Before use, the membranes were immersed for at least 12 h in a solution of the appropriate concentration.
2.2. Electrochemical characterization
Salt diffusion, impedance spectroscopy, and membrane po-tential measurements were carried out in a dead-end test cell similar to that described elsewhere[20]. The membranes were tightly clamped between two glass half-cells using silicone rub-ber rings. To minimize concentration polarization at the mem-brane surfaces, a magnetic stirrer was placed at the bottom of each half-cell and its speed rate was externally controlled; mea-surements were carried out at a stirring rate of 525 rpm.
• In salt diffusion measurements the membranes were ini-tially separating a concentrated solution (c1) from a diluted one (initially distilled water, c2=0). Changes in the so-lution on side 2 were recorded versus time by means of a conductivity cell connected to a digital conductivity me-ter (Radiomeme-ter CDM 83); a conductivity cell was also placed in reservoir 1 to control the constancy of concentra-tionc1. To see the effect of concentration gradients on salt permeability, two concentrated solutions (c1=10−2 and 5×10−2M) were used.
• The electromotive force (E) between the two sides of the membranes caused by a concentration gradient was mea-sured by two reversible Ag/AgCl electrodes connected to a digital voltmeter (Yokohama 7552, 1 G input resis-tance). The electrodes were in contact with the solutions and the cell configuration was Ag/AgCl electrode/NaCl solution (c1)/membrane/NaCl solution (c2)/Ag/AgCl elec-trode. Measurements were carried out by keeping the
con-centration of the solution at one side of the membrane constant (c1=cc=0.01 M) and changing gradually the concentration of the solution at the other side (c2=cv) from 0.01 to 0.1 M. In all cases the concentration c2 was in contact with the real membrane surface, which means the polyamide sublayer for composite PAC samples and the polysulfone for the ultrafiltration PSU ones; the electrode in contact with the constant concentration was grounded (E=Ev−Ec). Membrane potential, Φm, was determined from measured E values by subtract-ing the electrode potential contribution,Φelec: Φm=
E−Φelec.
• Impedance spectroscopy (IS) measurements were carried out with an impedance analyzer (Solartron 1260) controlled by a computer. The experimental data were corrected by software as well as the influence of connecting cables and other parasite capacitances. The measurements were car-ried out using 100 different frequencies in the range 10– 107Hz, at a maximum voltage of 0.01 V, the solutions at both sides of the membrane having the same concentration.
2.3. X-ray photoelectron spectroscopy measurements
Surface chemical characterization of the different membrane samples was carried out by X-ray photoelectron spectroscopy (XPS) spectra, which were obtained using a Physical Elec-tronics PHI 5700 spectrometer with nonmonochromatic MgKα
radiation as excitation source (300 W, 15 kV, 1253.6 eV). High-resolution spectra were recorded at a given take-off an-gleφ=45◦by a concentric hemispherical analyzer operating in the constant-pass-energy mode at 29.35 eV, using a 720-µm-diameter analysis area. Under these conditions the Au 4f7/2 line was recorded with 1.16 eV full width at half maximum (FWHM) at a binding energy of 84.0 eV [21]; charge refer-encing was done by setting the C–C peak at 284.6 eV. The pressure in the analysis chamber was maintained lower than 5×10−6Pa. Membranes were irradiated for a maximum time of 15–20 min to minimize X-ray induced sample damage[22]. A PHI ACCESS ESCA-V6.0 F software package was used for acquisition and data analysis [23]. Atomic concentration per-centages of the membrane surface (polyamide) characteristic elements were determined from the measured spectral regions by taking into account the corresponding area sensitivity fac-tor[24].
3. Results and discussion
3.1. XPS study
Prior to transport characterization of the membranes, a chem-ical study of the polyamide top layer of the composite PAC-X samples was carried out by XPS analysis in order to estab-lish that there was no damage associated with the treatment with HNO3. Assuming the depth studied by XPS is around 6– 7 nm in polymeric samples [21], while the thickness of the polyamide active layer for fresh membranes ranged between 0.1 and 1 µm[1], only the polyamide top layer should be considered
Table 1
Effect of HNO3 treatment time on the atomic concentration percentages
(A.C. %) of the element present on the active surface (polyamide) of the com-posite PAC-X membrane samples and element ratios
Membrane C1s(%) O1s(%) N1s(%) S1s(%) N/C N/O PAC-F 74.8 14.0 11.0 0.2 0.147 0.786 PAC-0 75.6 13.0 11.2 0.2 0.148 0.862 PAC-12 68.5 21.5 9.7 0.3 0.142 0.451 PAC-24 69.0 21.3 9.3 0.4 0.135 0.437 PAC-48 72.2 20.8 6.6 0.4 0.093 0.324 PAC-72 74.9 18.0 6.8 0.3 0.091 0.383
in the analysis of the PAC-X samples. Relative atomic concen-trations (A.C. %) of the elements present on the surface of the PAC-X samples as well as the N/C and N/O molar ratio are in-dicated inTable 1. As can be observed, rather good agreement between experimental and theoretical values for the atomic concentration ratios of the nontreated samples was obtained ([N/C]T =0.143 and [N/O]T=1), but both ratios decreased when chemically treated samples were considered (polyamide oxidation). Small concentrations (less than 0.4%) of other non-characteristic polyamide elements (Cl, Na, S, or Fe) were also found in some of the samples, which are usually considered as impurities due to environment contamination or residual prod-ucts from membrane manufacturing[25]. In the case of sulfur, which is a characteristic element of the polysulfone support, the small concentrations obtained in all the PAC-X samples are also indicated inTable 1, since an increase in sulfur concentration with treatment time could indicate damage to the polyamide top layer as a consequence of nitric acid treatment, but that effect was not observed.
Fig. 1shows a comparison of C1s, O1s, and N1sspectra for PAC-0 and PAC-48 samples, and some differences can be ob-served when carbon and oxygen spectra are compared, but the N1sspectra are practically independent of the acidic treatment. C1s spectra were fitted by considering three different carbon contributions[26–28]: (i) aliphatic/aromatic (plus adventitious carbon) at 284.8 eV B.E. (carbon CA); (ii) the peak at 286.1 eV B.E. corresponding to –C–O– or –C–N– (carbon CB); (iii) the peak at 287.9 eV B.E. due to C=N or O=C–N (carbon CC). Two different contributions were considered by fitting oxygen and nitrogen spectra[29,30]: (i) oxygen OA, with the peak at 532.0 eV for C=O, O=C–N; (ii) the peak at 533.5 eV asso-ciated with H≡O=C–N and named OB; (iii) the nitrogen at 399.2 eV corresponding to –C=N– (nitrogen NA) and the peak of nitrogen NB, for O=C–N, appearing at 400.4 eV[31].Fig. 2
shows experimental and fitted signals corresponding to C1sand O1s spectra for the PAC-0 membrane, andTable 2shows the percentages of the total area assigned to each contribution for the different PAC-X membranes. As can be observed, the ni-trogen percentage is practically independent of treatment time (except for the PAC-72 sample).
The main information obtained from XPS analysis of the PAC-X surfaces after treatment with 1 M HNO3solution at dif-ferent times can be summarized in the following points: (i) no peak at high binding energy associated with the presence of ni-trate was observed; (ii) the increase in the percentage of oxygen
Fig. 1. XPS spectra for polyamide characteristic elements for two samples: PAC-0 (!) and PAC-48 (P).
Fig. 2. Deconvolution of C1sand O1ssignals for sample PAC-0: carbon CA(- - -), carbon CB(· · ·), and carbon CC(·-·); oxygen OA(- - -) and oxygen OB(· · ·).
and the clear increase in the intensity of the C1speaks at 286.5 and 287.9 eV assigned to oxidized carbon species indicates the partial oxidation of the active polysulfone top layer; (iii) the increase in the atomic concentration of carbon after 48 h of treatment could be due to the partial detection of the polysul-fone sublayer, which contains a higher percentage of carbon. According to these results, the nitric acid treatment of PAC sam-ples produce the oxidative degradation of the polyamide active layer, which depends on the treatment time, but no nitration of the back bone polymer seems to occur.
3.2. Salt permeability
Salt permeability through a membrane can be determined from salt diffusion measurements. According to Fick’s first law, the molar salt flux through a membrane (for a quasi-steady state) due to a concentration difference can be written as
(1)
Js=Ps(c1−c2)=(1/Sm)(dn/dt )=(V0/Sm)(dc2/dt ), whereJs is the diffusive salt flux and Ps is the salt or diffu-sional permeability in the membrane;c1andc2are the external
Table 2
Effect of HNO3treatment time on the concentration percentages of the different
types of carbon (CA, CB, and CC), oxygen (OAand OB), and nitrogen (NA
and NB) present on the active surface (polyamide) of the composite PAC-X
membrane samples
Membrane CA(%) CB(%) CC(%) OA(%) OB(%) NA(%) NB(%)
PAC-F 69.8 14.9 15.3 28.8 71.2 21.1 78.9 PAC-0 71.2 14.1 14.7 29.2 70.8 21.5 78.5 PAC-12 64.3 16.2 19.5 48.3 51.7 28.3 71.5 PAC-24 63.6 16.6 19.9 35.6 64.4 31.6 68.4 PAC-48 64.5 17.6 17.9 35.4 64.6 29.3 70.7 PAC-72 76.9 10.7 12.4 56.8 43.2 37.5 62.5
Fig. 3. Variation in the conductivity of solution 2 versus time.c1=0.01 M
NaCl. (P) PSU-24, (Q) PAC-24, (E) PSU-72, (F) PAC-72.
concentrations; V0 is the volume of the solution at the side of concentrationc2; andSm represents the membrane area. If concentrationc1can be considered constant, the following ex-pression is obtained,
(2)
(dσ2/dt )=(S/V0)(dσ/dc)ePsc,
where(dσ/dc)eis an electrolyte characteristic parameter (for a given temperature).Fig. 3shows the variation in the conductiv-ity of the solutionc2as a function of time for some of the stud-ied samples, at the same constant concentration (c1=0.01 M). According to Eq.(2), salt permeability through the membranes can be determined from the slopes of the straight lines shown in Fig. 3. These results are indicated in Table 3, where the effect of age and the time of HNO3 treatment on Ps values can be observed; for comparison, salt permeability for fresh
PSU and PAC membranes are also presented. The results show: (a) a clear decrease in salt permeability as a result of aging for PAC and PSU samples, which is mainly attributed to the col-lapse of the porous structure, which is less significant in the case of the composite membrane, probably due to a partial protection by the polyamide top layer; (b) the effect of HNO3treatment in increasing salt permeability for both membranes and the de-termination of an optimum time (24 h for PSU samples and between 12 and 48 h for PAC). This point will be discussed in connection with the electrical resistance results to be presented inTable 5, where the contribution of each sublayer to the values obtained for the composite membrane can be estimated. On the other hand, the reduction in salt permeability when the concen-tration of NaCl increases obtained for the different membrane samples can also be observed in the results presented inTable 3.
3.3. Membrane potential results
The transport of ions across a membrane is studied by mea-suring the membrane potential (Φm), which is the electrical potential difference at both sides of a membrane separating two solutions of the same electrolyte but different concentrations (c1 andc2). According to the Teorell–Meyer–Sievers or TMS theory[32,33], the membrane potential can be considered as the sum of two Donnan potentials (one at each membrane/solution interface) plus a diffusion potential in the membrane, this means thatΦm=ΦDon(I)+Φdif+ΦDon(II). However, for
neu-tral or weakly charged membranes (where the salt concentration is much higher than the membrane fixed charge), the Donnan potential can be neglected andΦm is considered as a diffu-sion potential[34]. For 1:1 electrolytes, it is expressed by
(3)
Φm≈Φdif=(RT /F )(1−2t+)ln(a1/a2),
wherea1anda2are the mean ionic activities of the solutions at the two sides of the membrane (in the case of diluted solu-tions concentrasolu-tions can be used instead of activities),t+ and
t−are the cation and anion transport numbers, respectively;R,
F, andT represent the gas constant, Faraday constant, and the temperature of the system. The transport number of the ioniin the membrane represents the amount of current transported for one ion with respect to the total current crossing the membrane,
ti=Ii/IT, and for single saltst++t−=1.
Variation of membrane potential with salt activity ratio for the different studied membranes is shown in Fig. 4. As can
Table 3
Effect of age and HNO3treatment time on salt permeability through the polysulfone porous support (PSU-X) and the composite polyamide/polysulfone membranes
(PAC-X) determined at two different NaCl constant concentrations (0.01 and 0.05 M)
Membrane c=0.01 M c=0.05 M Membrane c=0.01 M c=0.05 M
Ps(m/s) Ps(m/s) Ps(m/s) Ps(m/s)
PSU-F 170.0×10−9 130.0×10−9 PAC-F 40.0×10−9 30.0×10−9
PSU-0 5.1×10−9 0.9×10−9 PAC-0 14.0×10−9 7.7×10−9
PSU-12 6.5×10−9 1.1×10−9 PAC-12 26.3×10−9 11.3×10−9
PSU-24 23.1×10−9 11.7×10−9 PAC-24 22.1×10−9 12.5×10−9
PSU-48 13.1×10−9 3.5×10−9 PAC-48 23.2×10−9 10.8×10−9
(a)
(b)
Fig. 4. Membrane potential versus ln(av/af). (a) (1) PSU-0, (×) PSU-F, (!) PSU-12, (P) PSU-24, (e) PSU-48, and (E) PSU-72; (b) (1) PAC-0, (×) PAC-F, (!) PAC-12, (P) PAC-24, (e) PAC-48, and (E) PAC-72.
be observed, linear Φm– ln(a1/a2) relationships were ob-tained for all the samples and the cation transport number in the membranes, for each pair of external solutions, was de-termined from Eq.(3). Average value of the cation transport number for the whole interval of concentration studied, t+, and membrane is indicated inTable 4. As can be seen, t+
values for the different membranes hardly differ from each other (and from the solution value,t+0 =0.385), in agreement with the almost neutral character of these kind of membranes. These results indicate that age and chemical treatment do not cause electrical changes in the membrane, and the opening of the membrane structure assumed from diffusion results does not affect the transport numbers, since they represent relative fluxes.
3.4. Impedance spectroscopy
To correlate the modifications due to the effect of aging and chemical treatments in the whole composite membrane with
Table 4
Effect of age and HNO3treatment time on the average cation transport
num-ber,t+, through the polysulfone porous support (PSU-X) and the composite polyamide/polysulfone (PAC-X) membrane samples
Membrane t+ Membrane t+
PSU-F (0.393±0.012) PAC-F (0.415±0.012)
PSU-0 (0.386±0.018) PAC-0 (0.423±0.023)
PSU-12 (0.399±0.024) PAC-12 (0.395±0.011)
PSU-24 (0.393±0.021) PAC-24 (0.392±0.013)
PSU-48 (0.376±0.019) PAC-48 (0.388±0.016)
PSU-72 (0.384±0.011) PAC-72 (0.394±0.017)
(a)
(b)
Fig. 5. Impedance plots for an electrolyte solution: (a) Nyquist plot; (b) Bode plot. Experimental values (×), calculated values (–2–), assuming a parallel R–C circuit (R=16.800andC=4×10−12F).
those corresponding to each sublayer, impedance spectroscopy measurements were performed. The analysis of impedance data is usually carried out by the complex planeZ∗ method, which involves plotting the impedance imaginary part,−Zimg, versus the real part, Zreal (Nyquist plot). A single parallel R–C cir-cuit gives rise to a semicircle in the Z∗ plane, as shown in
Fig. 5a for an electrolyte solution placed between two elec-trodes (c=0.002 M NaCl), which has intercepts on the Zreal axis atR∞(ω→ ∞) and R0 (ω→0), being(R0−R∞)the resistance of the system. The maximum of the semicircle equals 0.5(R0−R∞)and it occurs at such a frequency thatωRC=1,
RC=τ being the relaxation time andωthe angular frequency (ω=2πf)[35]. Real and imaginary parts of the impedance are related to the circuit parametersRandC by the following ex-pressions:
Zreal=R/
1+(ωRC)2
and Zimg= −ωCR2/
1+(ωRC)2.
A comparison between experimental and calculated values (R=16.800andC=4×10−12F) is also shown inFig. 5. The Bode plot (−Zimgversus frequency) allows the determina-tion of the maximum frequency associated to a given relaxadetermina-tion process; for the electrolyte solution previously considered, the Bode plot presents a unique relaxation process with a maximum frequencyfmax≈2 MHz, as can be observed inFig. 5b.
However, complex systems usually present a distribution of relaxation times and the resulting plot is a depressed semicircle; in such cases a nonideal capacitor, or constant phase element (CPE), can be considered [35]. The impedance for the CPE is expressed by Q(ω)=Y0(ω)−n, where the admittance Y0 (s−n) andnare two empirical parameters (0n1). A par-ticular case is obtained whenn=0.5; then the circuit element corresponds to a “Warburg impedance” (W), which is associ-ated with a diffusion process according to Fick’s first law. On the other hand, in the case of heterogeneous systems consist-ing of a series array of layers with different electrical properties (such as electrolyte/membrane systems), the equivalent circuits obtained by the impedance data allow the separate determina-tion of the electrical parameters associated to each sublayer.
Nyquist and Bode plots obtained with PSU and PAC mem-brane systems are shown in Figs. 6 and 7, respectively (c=
2×10−3M NaCl), and, as can be observed, in all cases the con-tribution of the electrolyte solution is clearly differentiated from that corresponding to the membranes. Two relaxation processes can be seen in both kinds of diagrams for PSU-X/electrolyte systems (Fig. 6), which correspond to the membrane itself (m) and the electrolyte solution placed between the membrane sur-face and the electrodes (e); the equivalent circuit associated to these systems consists in a series association of two sub-circuits:
(i) The membrane contribution, which consists in a parallel as-sociation of a resistance and a nonideal capacitor (RmQm); (ii) The electrolyte contribution, given by a parallel association
of a resistor and a capacitor (ReCe).
However, three relaxation processes can be observed in
Fig. 7for the composite PAC-X membrane/electrolyte systems (the porous layer (p.l.) and the active layer (a.l.) of the com-posite membrane are obtained separately), and the equivalent circuit is represented by the following contributions:
(i) A parallel association of a resistor and a capacitor, (RaCa), for the dense active layer;
(ii) A parallel association of a resistor and a nonideal capac-itor, (RpQp), for the porous sublayer, according to the equivalent circuit previously determined for the porous PSU membranes;
(iii) The electrolyte solution, represented by (ReCe).
Differences among the same kind of samples depending on the time of treatment with HNO3can also be observed inFigs. 6
(a)
(b)
Fig. 6. Impedance plots for PSU-X samples at a given NaCl concentration: (a) Nyquist plot; (b) Bode plot. (1) PSU-0, (!) PSU-12, (P) PSU-24, (×) PSU-48, and (E) PSU-72.
and 7, which are particularly significant for the PSU-24 sam-ple; the shift to a higher value of the maximum frequency (fmax=10 kHz) and the decrease in the electrical resistance of the PSU-24 membrane indicates a more open structure, which is in agreement with the higher salt permeability obtained from diffusion results for this sample.
The fitting of the impedance experimental points by means of a nonlinear program allows the determination of the different circuit parameters[36].Figs. 8a and 8bshow the concentration dependence for the resistance values assigned to the porous and active sublayers of PAC-X samples (Ra andRp, respectively), while Fig. 9 shows the electrical resistance for PSU-X sam-ples (Rm); for comparison, values determined for the porous sublayer of two PAC samples (Rp values) are also indicated inFig. 9. In all cases, a decrease in the values of the electri-cal resistance when the salt concentration increases was found, which is associated to the increase of the conductivity of the electrolyte solution embedded in the membrane matrix[18,37].
Table 5 shows a comparison of the average electrical re-sistance ratios for the different samples (RPSUm -X/RPSUm -F, RPAC-X
p /RPACp -F, andRaPAC-X/RPACa -F); the value of each av-erage resistance ratio was obtained as the mean of six values
(a)
(b)
Fig. 7. Impedance plots for some PSU-X samples at a given NaCl concentra-tion (0.002 M): (a) Nyquist plot; (b) Bode plot. (!) PSU-12, (×) PSU-24, and (e) PSU-48.
corresponding to the ratio of the electrical resistance for a given membrane (or membrane sublayer) at a particular NaCl concen-tration with respect to the value of the fresh sample/sublayer for the same salt solution. Since no electrical modification in the membrane matrix is assumed according to membrane potential results, variations in the values of the electrical resistance ratio are attributed to modifications in the membrane structure. As can be observed, age strongly increases the electrical resistance of PSU samples, which seems to be associated to changes in the polysulfone matrix caused by the loss of humidic compounds (polymer chains compaction), but nitric acid treatment for 24 h clearly reduces that effect. Moreover, results for the PAC com-posite polyamide/polysulfone membranes show a much lower increase in the electrical resistance associated to the porous polysulfone sublayer (Rp), which could be due to the presence of the polyamide top layer covering the polysulfone sublayer (and the unwoven support on the other side), which would di-minish the lost of humidic substances, reducing the compaction of the porous sublayer in comparison with the obtained for the ultrafiltration membrane. A similar effect of HNO3 treat-ment was already observed with diffusion measuretreat-ments (in that case, higher diffusional permeability was obtained, as can be seen inTable 3), which is in agreement with results reported
(a)
(b)
Fig. 8. Variation of the electrical resistance of the active (a) and porous (b) sublayers of the PAC-X samples with NaCl concentration. PAC-0 ("), PAC-12 (E), PAC-24 (!), PAC-48 (P), PAC-72 (e), PAC-F (×).
Fig. 9. Variation of the electrical resistance,Rm, of the PSU-X samples with NaCl concentration. PSU-0 ("), PSU-12 (E), PSU-24 (!), PSU-48 (P), PSU-72 (e), PSU-F (×), and porous sublayer of PAC-F sample (1).
in the literature (increase of hydrodynamic permeability after treatment with HF and its time dependence[38]). In any case, the different interactions between the acidic solution and the ultrafiltration membrane or the porous sublayer of the compos-ite samples might also explain the differences observed among them. It should be pointed out that impedance spectroscopy re-sults are in good agreement with that obtained from diffusion measurements but they allow a more specific assignation of age
Table 5
Effect of age and HNO3 treatment time on the average electrical
resis-tance ratio for each PSU-X sample with respects to the fresh membrane,
RPSU-Xm /RmPSU-F, and the porous and active sublayers,RpPAC-X/RPAC-Fp and RPAC-Xa /RaPAC-F, for each composite PAC-X sample
Membrane RPSU-Xm /RmPSU-F RpPAC-X/RPAC-Fp RaPAC-X/RPAC-Fa
PSU-0 98±18 15±4 5.0±1.2
PSU-12 35±2 2.6±0.2 4.6±1.7 PSU-24 8±1 2.8±0.3 4.5±1.3 PSU-48 22±2 2.5±0.6 7.8±1.6 PSU-72 25±3 5.8±0.5 3.8±1.1
and chemical effects to the matrix of the active/porous sublay-ers of composite membranes.
4. Conclusions
Age strongly affects membrane permeability and electrical resistance of reverse osmosis composite polyamide/polysulfone membranes, but immersion of membrane samples in 1 M HNO3 solution (HNO3 treatment) reduces age effects. XPS results show the oxidative degradation of the polyamide active layer for PAC samples (but it does not means sublayer cracking), and no electrical changes seem to occur according to mem-brane potential results for treatment times up to 72 h. Thus, membrane changes should be related to modifications in the structure/geometry of active and/or porous membrane sublay-ers.
Impedance spectroscopy measurements permit the separate estimation of each sublayer contribution, and the comparison of IS results obtained for the composite membrane and ultrafiltra-tion membrane similar to that used as its support layer shows stronger age effect on the porous polysulfone sublayer than on the active polyamide layer.
This study also shows that 24-h immersion in HNO3 solu-tion is the optimum time for the polysulfone membrane, since it clearly reduces the effect of age increasing permeability and de-creasing electrical resistance, but similar values were obtained for both parameters for the composite membrane after 12–48-h treatment.
References
[1] M. Mulder, Basic Principles of Membrane Technology, Kluwer, Dor-drecht, 1992.
[2] M. Cheryan, Ultrafiltration Handbook, Tecnomic, Lancaster, PA, 1986. [3] O. Kedem, A. Katchalsky, Trans. Faraday Soc. 59 (1963) 1941. [4] G. Jonsson, J. Benavente, J. Membr. Sci. 69 (1992) 29. [5] J. Benavente, G. Jonsson, J. Membr. Sci. 172 (2000) 189. [6] G. Jonsson, J. Membr. Sci. 14 (1983) 211.
[7] R.R. Bhave, Inorganic Membranes: Synthesis, Characteristics and Appli-cations, Van Nostrand–Reinhold, New York, 1991.
[8] A.E. Childress, M. Elimelech, J. Membr. Sci. 119 (1996) 253. [9] M. Pontié, J. Membr. Sci. 154 (1999) 213.
[10] M. Ulbrich, K. Richau, H. Kamusewitz, Colloids Surf. A 138 (1998) 353. [11] M. Pontié, X. Chasseray, D. Lemordant, J.M. Lairé, J. Membr. Sci. 129
(1997) 125.
[12] J. Benavente, M.I. Vázquez, J. Colloid Interface Sci. 273 (2004) 547. [13] A. Dimov, M.I. Islam, J. Membr. Sci. 50 (1990) 97.
[14] D. Mukherjee, A. Kulkarni, W.N. Gill, Desalination 104 (1996) 239. [15] B. Malmgren-Hansen, T.S. Sorensen, J.B. Jensen, M. Hennenberg, J.
Col-loid Interface Sci. 130 (1989) 359. [16] A. Asaka, J. Membr. Sci. 50 (1990) 71.
[17] J. Benavente, J.R. Ramos-Barrado, A. Heredia, Colloids Surf. A 140 (1998) 333.
[18] J. Benavente, J.M. García, J.G. de la Campa, J. de Abajo, J. Membr. Sci. 114 (1996) 51.
[19] A. Cañas, M.J. Ariza, J. Benavente, J. Membr. Sci. 183 (2001) 135. [20] J. Benavente, A. Muñoz, A. Heredia, Solid State Ionics 97 (1997) 89. [21] D. Brigs, M.P. Seah, Practical Surface Analysis: Auger and X-Ray
Photo-electron Spectroscopy, Wiley, Chichester, England, 1995.
[22] M.J. Ariza, P. Prádanos, R. Rico, E. Rodriguez-Castellón, J. Benavente, Surf. Interface Anal. 35 (2003) 360.
[23] Multitechnique ESCA Software Reference Manual for the PC-ACCESS Software Version 6.0, Physics Electronics, Minneapolis, 1995.
[24] A.P. Pijpers, R.J. Meier, Chem. Soc. Rev. 28 (1999) 233.
[25] J.T.F. Keurentjes, J.G. Harbrecht, D. Brikman, J.H. Hanemaaijer, M.A. Cohen, H. van ’t Riet, J. Membr. Sci. 47 (1989) 333.
[26] G.P. López, D.G. Castner, B.D. Ratner, Surf. Interface Anal. 17 (1991) 267.
[27] K.M. Shakesherff, C. Evora, I. Soriano, R. Langer, J. Colloid Interface Sci. 185 (1997) 538.
[28] M.C. Davies, R.A.P. Lynn, J. Heran, A.J. Paul, J.C. Vickerman, J.F. Watts, Langmuir 11 (1995) 4313.
[29] D.T. Clark, A. Harrison, J. Polym. Sci. Polym. Chem. Ed. 19 (1981) 1945. [30] M.J. Ariza, E. Rodriguez-Castellón, R. Rico, J. Benavente, M. Muñoz, M.
Oleinikova, J. Colloid Interface Sci. 226 (2000) 151. [31] G.A. Schick, Z. Sun, Langmuir 10 (1994) 3105. [32] K.H. Meyer, J.F. Siever, Helv. Chim. Acta 19 (1936) 649. [33] T. Teorell, Discuss. Faraday Soc. 21 (1956) 9.
[34] N. Lakshminarayanaiah, Transport Phenomena in Membranes, Academic Press, New York, 1969.
[35] J.R. Macdonald, Impedance Spectroscopy, Wiley, New York, 1987. [36] B.A. Boukamp, Solid State Ionics 18/19 (1986) 136.
[37] J. Benavente, M. Oleinikova, M. Muñoz, M. Valiente, J. Electroanal. Chem. 451 (1998) 173.
