- Email: [email protected]
Synthesis and characterization of new ultrafiltration ceramic membranes for water treatment
Journal of Water Process Engineering xxx (xxxx) xxx–xxx
Contents lists available at ScienceDirect
Journal of Water Process Engineering journal homepage: www.elsevier.com/locate/jwpe
Synthesis and characterization of new ultrafiltration ceramic membranes for water treatment ⁎
A. Chouguia,b, , A. Belouateka, M. Rabiller-Baudryc a
SEA2M Laboratory, Abdelhamid ibn badis University of Mostaganem, 27000, Algeria Idn khaldoun University of Tiaret, 14000, Algeria c Rennes University, Instiute of Chemical Sciences of Rennes (UMR CNRS 6226), 263 General Leclerc Avenue, Rennes, 35000,France b
A R T I C LE I N FO
A B S T R A C T
Keywords: Membrane Dyes Zeta potential Electrostatic interactions Filtration Ceramic
The present work aims at developing and testing two new ultrafiltration ceramic membranes synthesized from a locally available barbotine, kaolin, zirconia (UM01 membrane) and 3-Mercaptopropyl-trimethyoxysilane (AUM01). The electrokinetic characterization of membranes highlighted a negative charge density over a wide range of pH. The addition of 3-Mercaptopropyl-trimethyoxysilane leads to a decrease in water permeability from around 95–35 L h−1 m−2 bar−1. For both membranes, it was recognized that salt and dye rejections were strongly influenced by electrostatic interactions with the membrane surface where a considerable loss of performance with the increasing of ionic solution strength was induced. The maximum dye rejections were obtained with the AUM01 membrane and were close to 100% and 80% for cationic (methyl green and neutral red) and anionic (reactive black5) dyes, respectively.
1. Introduction Altought the two thirds of the earth's surface are covered by water, only 3% of the world’s water is fresh water, and two-thirds of that is tucked is unavailable for our use. That is mainly caused by the presence of a variety of contaminants in raw water, where their average size ranges from the micrometric scale (e.g. bacteria) to a few tenths of nanometers (solvated ions). Membrane processes like microfiltration, ultrafiltration, nanofiltration and reverse osmosis have proved their efficiency in the field of water treatment in various industrial sectors such as drinking water production and industrial effluents treatment. Ceramic membranes offer better properties as compared to organic ones in term of thermal, chemical, and mechanical stability as well as resistance to microbial degradation [1]. Moreover, they are suitable for applications in many fields such as wastewater recycling, textile, food and pharmaceutical industries [2,3]. The electrokinetic properties of membranes play a significant role in their separation performance and fouling tendency, since they exert a high influence on the magnitude of the interactions between the membrane and the feed liquid, Thus, the permeation fluxes of both solvent and solutes is affected through the membrane pores [4,27]. Ceramic membranes electrokinetic properties are commonly studied by electrophoretic mobility [5,6], electroosmosis flow rate [7] and
⁎
streaming potential measurements [6–12]. Membrane potential measurements can also be used to get relevant information about the membrane density charge [13,14]. Clays are interesting materials for ceramic supports and membranes. In our previous work a thin layer of alkoxide (98% tetraethylorthosilicate) was coated onto a kaolin/zirconia membrane and the resulting membrane showed interesting results for dyes and heavy metals removal [15,25]. In the present work, we synthesized new ultrafiltration ceramic membranes from clays, zirconia and alkoxide. These membranes were further characterized by X-ray fluorescence, X-ray diffraction spectroscopy, scanning electron microscopy, infrared spectroscopy and tangential streaming potential measurements. Finally, their performance in terms of salts and dyes rejection was investigated. 2. Materials and methods 2.1. Chemicals Kaolin used in this work was an Algerian natural product, a barbotine from the Ghazaouet area (western part of Algeria). Zirconium oxide (MW = 123.22 g/mol) and 3-Mercaptopropyl-trimethyoxysilane (MW = 196.34 g/mol, assay percent range 85%) were purchased from
Corresponding author at: SEA2M Laboratory, Abdelhamid ibn badis University of Mostaganem, 27000, Algeria. E-mail address: [email protected] (A. Chougui).
https://doi.org/10.1016/j.jwpe.2018.04.017 Received 3 August 2017; Received in revised form 11 April 2018; Accepted 26 April 2018 2214-7144/ © 2018 Elsevier Ltd. All rights reserved.
Please cite this article as: Chougui, A., Journal of Water Process Engineering (2018), https://doi.org/10.1016/j.jwpe.2018.04.017
Journal of Water Process Engineering xxx (xxxx) xxx–xxx
A. Chougui et al.
Salt rejection of Na2SO4, MgCl2, MgSO4 and NaCl solutions at pH 6.0 ± 0.2 and different concentrations (500, 1000 and 1500pm) was determined according to:
Sigma-Aldrich and Acros Organics, respectively. The feed solutions were prepared from analytical grade chemical products from Fluka (Methyl green, Neutral red and RB5) and Merck (Na2SO4, MgCl2, MgSO4, KCl and NaCl). Further, all solutions were prepared from milliQ quality water.
Cp ⎤ R (%) = 100 ⎡1 − ⎢ C F⎥ ⎦ ⎣
2.2. Membrane synthesis
Where cp and cF are the permeate and feed concentrations, respectively. Salt concentrations in the permeate solutions were determined from conductivity measurements. Eq. (2) was used to determine dye rejections too. Synthetic dye solutions (5 L) at three concentration (10−5, 10−4, 10−3 M) were filtered tangentially for 2 h under a TMP of 4 bar. Both permeate and feed concentrations were determined from visible spectrophotometry (optizen120 UV). The main characteristics of the dye solutions are collected in Table 1. All rejection results reported hereafter represent the average values of three successive experiments.
We synthesized two different kinds of membranes. A support layer made up of a barbotine: get from the western part of Algeria; was first prepared according to the protocol described in [16]. A layer consisting of kaolin and ZrO2(3%) was further coated onto the support. After sintering at 1050 °C for 2 h, a composite membrane (labeled UM01) composed of the support layer and a 10 μm thick skin layer with pore diameters around 0.1 μm was obtained. The second membrane (labeled AUM01) was obtained by adding a thin layer of alkoxide (3-Mercaptopropyl-trimethyoxysilane 85% (MPTMS) onto the UM01 membrane surface. This top layer was prepared via a sol–gel technique by mixing H2O: HCl(3%): MPTMS: Ethanol in the volume ration of 1.1: 1.1: 1: 1.1 from a colloidal suspension of MPTMS in alcohol followed by peptization under acidic conditions. The presence of the intermediate kaolin/ZrO2 layer is here essential since it connects as a “bridge” between the macroporous support and the top layer. Thus, it is of a great importance to avoid infiltration of this latter into the support when the transmembrane pressure (TMP) is applied. The slip-casting procedure was followed by drying and calcination at 780 °C.
3. Results and discussion 3.1. Membrane characterization 3.1.1. Chemical analysis The chemical analysis of the UM01 membrane was performed from X-ray fluorescence (Table 2). It indicated that SiO2 and Al2O3 were the major components of the powder. 3.1.2. XRD analysis The XRD spectra of the UM01 membrane skin layer obtained before and after sintering at 1050 °C are shown in Fig. 2(a) and (b), respectively. Before sintering, the main phases observed were kaolinite (K), muscovite (M), and quartz (Q). However, for samples sintered at 1050 °C, the kaolinite phase was no longer detected. XRD shows that mullite phase (Mu) was the main crystalline mineral in the sintered powder, and zirconia (Zr, monoclinic system) was detected as a minor phase. Moreover, the peaks observed for the AUM01 skin layer (Fig. 2(c)) at 2θ = 21° and 28° corresponding to Si and S peaks, respectively, clearly confirmed the coating of MPTMS over the membrane surface. During heating, from 950 °C metakaolinis transformed into aluminum-silicon spinel (Si3Al4O12), which is also referred to as a gammaalumina type structure. Upon calcination to around 1050 °C, the spinel phase nucleates and is transformed into mullite [3Al2O3·2SiO2], and highly crystalline cristobalite. Shrinkage of the sample volume is significant and porosity decreases [20,21].These identified phases are of great importance because of their interesting mechanical properties such as high Young’s modulus.
2.3. Membrane characterization The membrane chemical analysis was performed by X-ray fluorescence (ARL™ OPTIM'X WDXRF Spectrometer). X-Ray diffraction (XRD) was carried out with an XPert MPD automated powder diffractometer, using CuKa1 radiation source (a = 1.54056 Å). The structure and morphology of the two membrane were characterized by scanning electron microscopy (JEOL, JMC-6000). The infrared spectra of the MPTMS layer were obtained with a FT/IR-4100 Fourier Transform Infrared spectrometer (Jasco). Spectra were collected from 400 to 4000 cm−1 at 2 cm−1 resolution and each spectrum was averaged from 128 scans after background recording performed at ambient air. A Zeta Cad (CAD Inst., France) electrokinetic analyzer was used for tangential streaming potential (TSP) measurements. The device measures the electrical potential difference generated by the pressure-induced circulation of an electrolyte solution (a milli molar KCl solution was used in this work) through a thin slit channel formed by a couple of identical membranes. It is worth mentioning that TSP measurements were performed with flat-sheet membrane samples. TSP (ΔE) was further used to assess the membrane apparent zeta potential (ζ) from the Helmholtz-Smoluchowski equation [17]:
ζ ΔE = ε0 εr ηλ 0 ΔP
(2)
3.1.3. Morphology Fig. 3 shows the images of SEM of two ceramic membranes before and after coating with MPTMS. There are any difference found between ceramic membranes before and after coating with thin layer. UM01 membrane before coating has large pore size compared with the AUM01 membrane. Ceramic membrane after coating with top layer has a smaller pore size because it has been covered by the MPTMS particles.
(1)
where ΔP is the pressure difference applied through the channel formed by the two membrane samples, ε0 is the vacuum permittivity, and εr, η and λ0 are the dielectric constant, viscosity and electric conductivity of the measuring solution, respectively.
3.1.4. FTIR spectroscopy analysis Fig. 4 presents the FTIR spectra of the AUM01 membrane MPTMS skin layer before (a) and after (b) sintering at 780 °C (see Section 2.2). FTIR spectra of MPTMS before sintering Fig. 4(a) reveal the large broad band appeared over the range of 3000–3400 cm−1 probably is emerge from presence of OeH. The peaks around 1100 and 1600 cm–1 may be due to vibrational and bending modes of HeOeH, respectively, and CeO stretching vibrations was appeared in 1000 cm−1 [26]. The band at 997 cm−1 and 1115 cm−1corresponds to the asymmetric
2.4. Ultrafiltration tests Tubular membranes (length: 30 cm; inner diameter: 13 mm; outer diameter: 16 mm) were used for cross-flow filtration experiments (a schematic of filtration unit is shown in Fig. 1) performed at a TMP of 4 bar. Ultrafiltration performance was mainly described through permeability (Lp), water flux (J) and rejection rate (R). 2
Journal of Water Process Engineering xxx (xxxx) xxx–xxx
A. Chougui et al.
Fig. 1. Cross-flow ultrafiltration unit. Table 1 Characteristics of the dyes solutions [18,19]. Dyes
Character
λmax(nm)
Initial pH
MW(g/mol)
Methyl green
Cationic
632
7.90
458.41
+/+
Neutral Red
Cationic
435
7.18
288.78
+
RB5
Anionic
599
7.02
991.82
-/-/-/-
Developed formula
Valence
around 3 and 2.8, respectively, the membrane surfaces being overall negatively charged for pH higher than these values and positively charged for lower pH. Such an amphoteric behavior comes from protonation/deprotonation of hydroxyl groups onto the membranes surfaces.The pH dependence of the UM01 and AUM01 zeta potentials depends upon the reactivity of surface hydroxyl groups with the potential determining ions H+, which modifies the net charge of the membrane surface. Although both membranes have relatively close i.e.p. the AUM01 membrane exhibits much larger (negative) zeta potentials than the UM01 membrane, which is expected to impact the membrane separation performance regarding charged solutes (due to stronger Donnan exclusion in the case of the AUM01 membrane).
stretching of the SieOeC and SieOeSi bonds, respectively [22,23]. The band at 1240 cm−1 is associated with SieReS stretching [22,23]. The FTIR spectrum recorded after sintering Fig. 4(b) shows the most prominent a new peaks was found around 1053 cm−1 that may be belong to the stretching and bending vibration of OeSieO and SieO of silanol groups band from MPTMS while the characteristic band due to OeH disappears after sintering. FTIR results suggest that MPTMS creates a SiO2 network by producing SieOeSi inter bonds. 3.1.5. Electrokinetic characterization Fig. 5 shows the pH dependence of zeta potential determined from TSP measurements for both UM01 and AUM01 membranes. The membranes surface charge density as well as their surface potential change with pH because the protonation degree of functional groups on the membrane surface is greatly dependent on pH [24]. The isoelectric points (i.e.p.) of the UM01 and AUM01 membranes were found to be
3.2. Water flux and permeability Pure water fluxes of both UM01 and AUM01 membranes are shown in Fig. 6. In both cases the water flux was found to vary linearly with
Table 2 Composition of the UM01powder determined from X-ray fluorescence. Oxide
SiO2
Al2O3
Fe2O3
TiO2
K2O
Na2O
CaO
MgO
ZrO2
pF
Raw(%) Baked (%)
47.96 54.16
34.46 38.92
0.87 0.98
< 0.243 < 0.274
1.505 1.7
< 0.097 < 0.109
< 0.097 < 0.438
< 0.388 < 0.109
3.00 3.29
11.46 –
3
Journal of Water Process Engineering xxx (xxxx) xxx–xxx
A. Chougui et al.
Fig. 2. XRD spectra of, (a): UM01 raw powder, (b): UM01 skin layer powder heated at 1050 °C for 2 h, (c): AUM01 skin layer (MPTMS).
Fig. 4. FT-IR spectra of the MPTMS layer (a) before and (b) after sintering at 780 °C.
Fig. 5. Variation of the membrane zeta potential with pH.
Fig. 3. Scanning electron microscopy of the UM01 membrane (a) and AUM01 membrane (b).
the TMP as expected from Darcy’s law. The hydraulic permeability deduced from the slope flux = f (TMP) indicated an almost three fold higher permeability for the UM01 membrane (88.4 L h−1 m−2 bar−1) than for the AUM01 membrane (31.7 L h−1 m−2 bar−1). A decrease in the flux was observed for both membranes during the first 30 min of operation before a steady-state was reached (Fig. 7). The same time was required to reach steady-state with salt and dye solutions. Lower flux obtained with salt solutions resulted from concentration polarization. Fig. 6. Water flux as a function of TMP.
4
Journal of Water Process Engineering xxx (xxxx) xxx–xxx
A. Chougui et al.
Fig. 7. Evolution of permeation flux through UM01 and AUM01 membranes over time for different feed solutions (TMP = 4 bar).
3.3. Retention rate of salts and dyes 3.3.1. Effect of concentration and ionic strength on salt rejection Salt rejection performance of both UM01 and AUM01 membranes obtained for single salt solutions (NaCl, MgCl2, Na2SO4 and MgSO4) in the pH range 5.8–6.2 and for a TMP of 4 bar is shown in Fig. 8. For both membranes and whatever, the electrolyte rejection rates were found to decrease with increasing feed concentrations (see Fig. 8(a) and (c)). This is the signature of a rejection mechanism ruled by the so-called Donnan exclusion. Salts with magnesium cations (MgCl2 and MgSO4) were less rejected than salts with sodium cations (NaCl, and Na2SO4), which is in line with negative zeta potentials measured for both membranes in the pH range 5.8–6.2 (see Fig. 5) since bivalent cations are more attracted by the negative membrane charge than monovalent cations. Fig. 8b and d confirm the dominant Donnan exclusion mechanism since a clear correlation between the salt rejection and the medium ionic strength was found. Indeed, an increase in the ionic strength leads to compression of the electrochemical double layer and then to weaker electrostatic repulsions. 3.3.2. Dye rejection Fig. 9 shows the variation of dye rejection as a function of concentration. The AUM01 membrane exhibits higher rejections than the UM01 membrane whatever the dye is. For the AUM01 membrane, rejection was found to be almost independent of the dye concentration (for the concentration range under investigation) while the UM01 membrane performance was more sensitive to dye concentration. This can be explained by the larger pores of the UM01 membrane as compared to the AUM01 membrane ones in addition to the weaker electrostatic interactions between charged dyes and the UM01 membrane. Interestingly, the rejection of the anionic dye, RB5, was lower than the one of cationic dyes, MG and NR although the pristine membranes were negatively charged (Fig. 5). A possible explanation is the adsorption of cationic dyes onto the negatively charged pores in the first times of filtration which causes a decrease in the pore size along the charge reversal surface. This leads to strong repulsive interactions between adsorbed and free MG and NR molecules. On the other hand, no (or at least limited) adsorption of RB5 is expected onto the like-charged membranes, which pore size is not reduced during RB5 filtration. 4. Conclusion
Fig. 8. Variation of salt rejection by UM01 and AUM01 membranes with salt concentration, (a) and (c), and ionic strength, (b) and (d).
The aim of this work was to develop, characterize and test new ultrafiltration ceramic membranes synthesized from a locally available barbotine, kaolin, zirconia and 3-Mercaptopropyltrimethyoxysilane. It was recognized from the tangential streaming potential measurements that membranes were negatively charged over a wide range of pH. Further, both salt and dye rejections were strongly influenced by
electrostatic interactions with the membrane surface. An anionic and two cationics were considered in ultrafiltration experiments. Very promising performance was obtained for the different dyes, with experimental rejections ranging from 80 to 100% for the 5
Journal of Water Process Engineering xxx (xxxx) xxx–xxx
A. Chougui et al.
References [1] A.J. Burggraaf, L. Cot, Fundamentals of inorganic membrane science and technology, Elsevier Science and Technology Series, No. 4, Elsevier, Amsterdam, 1996. [2] L. Xu, W.P. Li, S.Q. Lu, Z. Wang, Q.X. Zhu, Y. Ling, Treating dye waste water by ceramic membraneincrossflow microfiltration, Desalination 149 (2002) 199–203. [3] H. Zhang, X. Quan, S. Chen, H. Zhao, Y. Zhao, W. Li, Zirconia and titania composite membranesfor liquid phase separation: preparation and characterization, Desalination 190 (2006) 172–180. [4] J.-I. Oha, S.H. Lee, Influence of streaming potentialon flux decline of microfiltration with in-linerapid pre-coagulation process for drinking waterproduction, J. Membr. Sci. 254 (2005) 39–47. [5] L. Ricq, A. Pierre, J.-C. Reggiani, J. Pagetti, A. Foissy, Use of electrophoretic mobility and streamingpotential measurements to characterize electrokineticproperties of ultrafiltration and microfiltration membranes, Collids Surf. A 138 (1998) 291–299. [6] S. Condom, S. Chemlal, W. Chu, M. Persin, A. Larbot, Correlation between selectivity and surfacecharge in cobalt spinel ultrafiltration membrane, Sep. Purif. Technol. 25 (2001) 545–548. [7] A. Szymczyk, P. Fievet, M. Mullet, J.C. Reggiani, J. Pagetti, Comparison of two electrokineticmethods – electroosmosis and streaming potential — to determine the zeta-potential of plane ceramicmembranes, J. Membr. Sci. 143 (1998) 189–195. [8] Y. Elmarraki, M. Persin, J. Sarrazin, M. Cretin, A. Larbot, Filtration of electrolyte solutions withnew TiO2–ZnAl2O4 ultrafiltration membranes in relationwith the electric surface properties, Sep. Purif. Technol. 25 (2001) 493–499. [9] L. Ricq, A. Pierre, J.-C. Reggiani, J. Pagetti, Streaming potential and ion transmission during ultra-and microfiltration on inorganic membranes, Desalination 114 (1997) 101–109. [10] M. Nyström, M. Lindström, E. Matthiasson, Streaming potential as a tool in the characterization of ultrafiltration membranes, Collids Surf. A 36 (1989) 297–312. [11] P. Fievet, M. Sbaï, A. Szymczyk, C. Magnenet, C. Labbez, A. Vidonne, A new tangential streaming potential set-up for the electrokinetic characterization of tubular membranes, Sep. Sci. Technol. 39 (2004) 2931–2949. [12] A. Szymczyk, C. Labbez, P. Fievet, B. Aoubiza, C. Simon, Streaming potential through multilayer membranes, AIChE J. 47 (2001) 2349–2358. [13] P. Fievet, B. Aoubiza, A. Szymczyk, J. Pagetti, Membrane potential in charged porous membranes, J. Membr. Sci. 160 (1999) 267–275. [14] A. Szymczyk, P. Fievet, J.C. Reggiani, J. Pagetti, Characterisation of surface properties of ceramic membranes by streaming and membrane potentials, J. Membr. Sci. 146 (1998) 277–284. [15] A. Chougui, K. Zaiter, A. Belouatek, B. Asli, Heavy metals and color retention by a synthesized inorganic membrane, Arab. J. Chem. 7 (2014) 817–822. [16] A. Belouatek, N. Benderdouche, A. Addou, A. Ouagued, N. Bettahar, Preparation of inorganic supports for liquid waste treatment, Microporous Mesoporous Mater. 85 (2005) 163–168. [17] M. Smoluchowski, Zurtheorie der electrischenkataphorese und der oberfluchenleitung, Phys. Z 6 (1905) 529–536. [18] S.H. Christopher, B.M. Dennis, Some electrochemical and photoelectrochemical properties of 3-amino-7-dimethylamino-2-methylphenazine (neutral red) in aqueous solution, Aust. J. Chem. 36 (1983) 507–516. [19] S.K. Kim, B. Nordén, Methyl green: a DNA major-groove binding drug, FEBS Lett. 315 (1993) 61–64. [20] V. Hanykýř, J. Kutzendorfer, Technology of Ceramics, SilisPraha a Vega Hradec králové, Praha, 2000 (in Czech). [21] D.W. Breck, Zeolite Molecular Sieves, Wiley-Interscience, New York, 1974. [22] N. Hebalkar, S. Kharrazi, A. Ethiraj, J. Urban, R. Fink, S.K. Kulkarni, Structural and optical investigations of SiO2-CdS core-shell particles, J. Colloid Interface Sci. 278 (2004) 107–114. [23] B. Stuart, Modern Infrared Spectroscopy, John Wiley & Sons New York, NY, USA, 1996. [24] R. Takagi, M. Nakagaki, Characterization of the membrane charge of Al2O3 membranes, Sep. Purif. Technol. 25 (2001) 369–377. [25] K. Zaiter, A. Belouatek, A. Chougui, B. Asli, A. Szymczyk, Color, COD, and salt retention by inorganic membrane, Desalin. Water Treat. (2013) 1–7. [26] M. Kolaei, K. Dashtian, Z. Rafiee, M. Ghaedi, Ultrasonic-assisted magnetic solid hase extraction of morphine in urine samples by new imprinted polymer-supported on WCNT-Fe3O4-NPs: central composite design optimization, Ultrason. Sonochem. 33 (2016) 240–248. [27] M. Jamshidi, M. Ghaedi, K. Dashtian, S. Hajati, A. Bazrafshan, RSC Adv. 73 (2015) 59522–59532.
Fig. 9. Effects of dyes concentration on the retention.
most efficient membrane. Higher rejections were obtained with cationic dyes which were attributed to strong electrostatic repulsions between dyes and membranes after dye adsorption onto the membrane pores leading to (i) pore size reduction and (ii) surface charge reversal.
Acknowledgment We wish to thank gratefully Professor Anthony SZYMCZYK from the Institute of Chemical Sciences of Rennes who provided insight and expertise that greatly assisted the present manuscript.
6
Contents lists available at ScienceDirect
Journal of Water Process Engineering journal homepage: www.elsevier.com/locate/jwpe
Synthesis and characterization of new ultrafiltration ceramic membranes for water treatment ⁎
A. Chouguia,b, , A. Belouateka, M. Rabiller-Baudryc a
SEA2M Laboratory, Abdelhamid ibn badis University of Mostaganem, 27000, Algeria Idn khaldoun University of Tiaret, 14000, Algeria c Rennes University, Instiute of Chemical Sciences of Rennes (UMR CNRS 6226), 263 General Leclerc Avenue, Rennes, 35000,France b
A R T I C LE I N FO
A B S T R A C T
Keywords: Membrane Dyes Zeta potential Electrostatic interactions Filtration Ceramic
The present work aims at developing and testing two new ultrafiltration ceramic membranes synthesized from a locally available barbotine, kaolin, zirconia (UM01 membrane) and 3-Mercaptopropyl-trimethyoxysilane (AUM01). The electrokinetic characterization of membranes highlighted a negative charge density over a wide range of pH. The addition of 3-Mercaptopropyl-trimethyoxysilane leads to a decrease in water permeability from around 95–35 L h−1 m−2 bar−1. For both membranes, it was recognized that salt and dye rejections were strongly influenced by electrostatic interactions with the membrane surface where a considerable loss of performance with the increasing of ionic solution strength was induced. The maximum dye rejections were obtained with the AUM01 membrane and were close to 100% and 80% for cationic (methyl green and neutral red) and anionic (reactive black5) dyes, respectively.
1. Introduction Altought the two thirds of the earth's surface are covered by water, only 3% of the world’s water is fresh water, and two-thirds of that is tucked is unavailable for our use. That is mainly caused by the presence of a variety of contaminants in raw water, where their average size ranges from the micrometric scale (e.g. bacteria) to a few tenths of nanometers (solvated ions). Membrane processes like microfiltration, ultrafiltration, nanofiltration and reverse osmosis have proved their efficiency in the field of water treatment in various industrial sectors such as drinking water production and industrial effluents treatment. Ceramic membranes offer better properties as compared to organic ones in term of thermal, chemical, and mechanical stability as well as resistance to microbial degradation [1]. Moreover, they are suitable for applications in many fields such as wastewater recycling, textile, food and pharmaceutical industries [2,3]. The electrokinetic properties of membranes play a significant role in their separation performance and fouling tendency, since they exert a high influence on the magnitude of the interactions between the membrane and the feed liquid, Thus, the permeation fluxes of both solvent and solutes is affected through the membrane pores [4,27]. Ceramic membranes electrokinetic properties are commonly studied by electrophoretic mobility [5,6], electroosmosis flow rate [7] and
⁎
streaming potential measurements [6–12]. Membrane potential measurements can also be used to get relevant information about the membrane density charge [13,14]. Clays are interesting materials for ceramic supports and membranes. In our previous work a thin layer of alkoxide (98% tetraethylorthosilicate) was coated onto a kaolin/zirconia membrane and the resulting membrane showed interesting results for dyes and heavy metals removal [15,25]. In the present work, we synthesized new ultrafiltration ceramic membranes from clays, zirconia and alkoxide. These membranes were further characterized by X-ray fluorescence, X-ray diffraction spectroscopy, scanning electron microscopy, infrared spectroscopy and tangential streaming potential measurements. Finally, their performance in terms of salts and dyes rejection was investigated. 2. Materials and methods 2.1. Chemicals Kaolin used in this work was an Algerian natural product, a barbotine from the Ghazaouet area (western part of Algeria). Zirconium oxide (MW = 123.22 g/mol) and 3-Mercaptopropyl-trimethyoxysilane (MW = 196.34 g/mol, assay percent range 85%) were purchased from
Corresponding author at: SEA2M Laboratory, Abdelhamid ibn badis University of Mostaganem, 27000, Algeria. E-mail address: [email protected] (A. Chougui).
https://doi.org/10.1016/j.jwpe.2018.04.017 Received 3 August 2017; Received in revised form 11 April 2018; Accepted 26 April 2018 2214-7144/ © 2018 Elsevier Ltd. All rights reserved.
Please cite this article as: Chougui, A., Journal of Water Process Engineering (2018), https://doi.org/10.1016/j.jwpe.2018.04.017
Journal of Water Process Engineering xxx (xxxx) xxx–xxx
A. Chougui et al.
Salt rejection of Na2SO4, MgCl2, MgSO4 and NaCl solutions at pH 6.0 ± 0.2 and different concentrations (500, 1000 and 1500pm) was determined according to:
Sigma-Aldrich and Acros Organics, respectively. The feed solutions were prepared from analytical grade chemical products from Fluka (Methyl green, Neutral red and RB5) and Merck (Na2SO4, MgCl2, MgSO4, KCl and NaCl). Further, all solutions were prepared from milliQ quality water.
Cp ⎤ R (%) = 100 ⎡1 − ⎢ C F⎥ ⎦ ⎣
2.2. Membrane synthesis
Where cp and cF are the permeate and feed concentrations, respectively. Salt concentrations in the permeate solutions were determined from conductivity measurements. Eq. (2) was used to determine dye rejections too. Synthetic dye solutions (5 L) at three concentration (10−5, 10−4, 10−3 M) were filtered tangentially for 2 h under a TMP of 4 bar. Both permeate and feed concentrations were determined from visible spectrophotometry (optizen120 UV). The main characteristics of the dye solutions are collected in Table 1. All rejection results reported hereafter represent the average values of three successive experiments.
We synthesized two different kinds of membranes. A support layer made up of a barbotine: get from the western part of Algeria; was first prepared according to the protocol described in [16]. A layer consisting of kaolin and ZrO2(3%) was further coated onto the support. After sintering at 1050 °C for 2 h, a composite membrane (labeled UM01) composed of the support layer and a 10 μm thick skin layer with pore diameters around 0.1 μm was obtained. The second membrane (labeled AUM01) was obtained by adding a thin layer of alkoxide (3-Mercaptopropyl-trimethyoxysilane 85% (MPTMS) onto the UM01 membrane surface. This top layer was prepared via a sol–gel technique by mixing H2O: HCl(3%): MPTMS: Ethanol in the volume ration of 1.1: 1.1: 1: 1.1 from a colloidal suspension of MPTMS in alcohol followed by peptization under acidic conditions. The presence of the intermediate kaolin/ZrO2 layer is here essential since it connects as a “bridge” between the macroporous support and the top layer. Thus, it is of a great importance to avoid infiltration of this latter into the support when the transmembrane pressure (TMP) is applied. The slip-casting procedure was followed by drying and calcination at 780 °C.
3. Results and discussion 3.1. Membrane characterization 3.1.1. Chemical analysis The chemical analysis of the UM01 membrane was performed from X-ray fluorescence (Table 2). It indicated that SiO2 and Al2O3 were the major components of the powder. 3.1.2. XRD analysis The XRD spectra of the UM01 membrane skin layer obtained before and after sintering at 1050 °C are shown in Fig. 2(a) and (b), respectively. Before sintering, the main phases observed were kaolinite (K), muscovite (M), and quartz (Q). However, for samples sintered at 1050 °C, the kaolinite phase was no longer detected. XRD shows that mullite phase (Mu) was the main crystalline mineral in the sintered powder, and zirconia (Zr, monoclinic system) was detected as a minor phase. Moreover, the peaks observed for the AUM01 skin layer (Fig. 2(c)) at 2θ = 21° and 28° corresponding to Si and S peaks, respectively, clearly confirmed the coating of MPTMS over the membrane surface. During heating, from 950 °C metakaolinis transformed into aluminum-silicon spinel (Si3Al4O12), which is also referred to as a gammaalumina type structure. Upon calcination to around 1050 °C, the spinel phase nucleates and is transformed into mullite [3Al2O3·2SiO2], and highly crystalline cristobalite. Shrinkage of the sample volume is significant and porosity decreases [20,21].These identified phases are of great importance because of their interesting mechanical properties such as high Young’s modulus.
2.3. Membrane characterization The membrane chemical analysis was performed by X-ray fluorescence (ARL™ OPTIM'X WDXRF Spectrometer). X-Ray diffraction (XRD) was carried out with an XPert MPD automated powder diffractometer, using CuKa1 radiation source (a = 1.54056 Å). The structure and morphology of the two membrane were characterized by scanning electron microscopy (JEOL, JMC-6000). The infrared spectra of the MPTMS layer were obtained with a FT/IR-4100 Fourier Transform Infrared spectrometer (Jasco). Spectra were collected from 400 to 4000 cm−1 at 2 cm−1 resolution and each spectrum was averaged from 128 scans after background recording performed at ambient air. A Zeta Cad (CAD Inst., France) electrokinetic analyzer was used for tangential streaming potential (TSP) measurements. The device measures the electrical potential difference generated by the pressure-induced circulation of an electrolyte solution (a milli molar KCl solution was used in this work) through a thin slit channel formed by a couple of identical membranes. It is worth mentioning that TSP measurements were performed with flat-sheet membrane samples. TSP (ΔE) was further used to assess the membrane apparent zeta potential (ζ) from the Helmholtz-Smoluchowski equation [17]:
ζ ΔE = ε0 εr ηλ 0 ΔP
(2)
3.1.3. Morphology Fig. 3 shows the images of SEM of two ceramic membranes before and after coating with MPTMS. There are any difference found between ceramic membranes before and after coating with thin layer. UM01 membrane before coating has large pore size compared with the AUM01 membrane. Ceramic membrane after coating with top layer has a smaller pore size because it has been covered by the MPTMS particles.
(1)
where ΔP is the pressure difference applied through the channel formed by the two membrane samples, ε0 is the vacuum permittivity, and εr, η and λ0 are the dielectric constant, viscosity and electric conductivity of the measuring solution, respectively.
3.1.4. FTIR spectroscopy analysis Fig. 4 presents the FTIR spectra of the AUM01 membrane MPTMS skin layer before (a) and after (b) sintering at 780 °C (see Section 2.2). FTIR spectra of MPTMS before sintering Fig. 4(a) reveal the large broad band appeared over the range of 3000–3400 cm−1 probably is emerge from presence of OeH. The peaks around 1100 and 1600 cm–1 may be due to vibrational and bending modes of HeOeH, respectively, and CeO stretching vibrations was appeared in 1000 cm−1 [26]. The band at 997 cm−1 and 1115 cm−1corresponds to the asymmetric
2.4. Ultrafiltration tests Tubular membranes (length: 30 cm; inner diameter: 13 mm; outer diameter: 16 mm) were used for cross-flow filtration experiments (a schematic of filtration unit is shown in Fig. 1) performed at a TMP of 4 bar. Ultrafiltration performance was mainly described through permeability (Lp), water flux (J) and rejection rate (R). 2
Journal of Water Process Engineering xxx (xxxx) xxx–xxx
A. Chougui et al.
Fig. 1. Cross-flow ultrafiltration unit. Table 1 Characteristics of the dyes solutions [18,19]. Dyes
Character
λmax(nm)
Initial pH
MW(g/mol)
Methyl green
Cationic
632
7.90
458.41
+/+
Neutral Red
Cationic
435
7.18
288.78
+
RB5
Anionic
599
7.02
991.82
-/-/-/-
Developed formula
Valence
around 3 and 2.8, respectively, the membrane surfaces being overall negatively charged for pH higher than these values and positively charged for lower pH. Such an amphoteric behavior comes from protonation/deprotonation of hydroxyl groups onto the membranes surfaces.The pH dependence of the UM01 and AUM01 zeta potentials depends upon the reactivity of surface hydroxyl groups with the potential determining ions H+, which modifies the net charge of the membrane surface. Although both membranes have relatively close i.e.p. the AUM01 membrane exhibits much larger (negative) zeta potentials than the UM01 membrane, which is expected to impact the membrane separation performance regarding charged solutes (due to stronger Donnan exclusion in the case of the AUM01 membrane).
stretching of the SieOeC and SieOeSi bonds, respectively [22,23]. The band at 1240 cm−1 is associated with SieReS stretching [22,23]. The FTIR spectrum recorded after sintering Fig. 4(b) shows the most prominent a new peaks was found around 1053 cm−1 that may be belong to the stretching and bending vibration of OeSieO and SieO of silanol groups band from MPTMS while the characteristic band due to OeH disappears after sintering. FTIR results suggest that MPTMS creates a SiO2 network by producing SieOeSi inter bonds. 3.1.5. Electrokinetic characterization Fig. 5 shows the pH dependence of zeta potential determined from TSP measurements for both UM01 and AUM01 membranes. The membranes surface charge density as well as their surface potential change with pH because the protonation degree of functional groups on the membrane surface is greatly dependent on pH [24]. The isoelectric points (i.e.p.) of the UM01 and AUM01 membranes were found to be
3.2. Water flux and permeability Pure water fluxes of both UM01 and AUM01 membranes are shown in Fig. 6. In both cases the water flux was found to vary linearly with
Table 2 Composition of the UM01powder determined from X-ray fluorescence. Oxide
SiO2
Al2O3
Fe2O3
TiO2
K2O
Na2O
CaO
MgO
ZrO2
pF
Raw(%) Baked (%)
47.96 54.16
34.46 38.92
0.87 0.98
< 0.243 < 0.274
1.505 1.7
< 0.097 < 0.109
< 0.097 < 0.438
< 0.388 < 0.109
3.00 3.29
11.46 –
3
Journal of Water Process Engineering xxx (xxxx) xxx–xxx
A. Chougui et al.
Fig. 2. XRD spectra of, (a): UM01 raw powder, (b): UM01 skin layer powder heated at 1050 °C for 2 h, (c): AUM01 skin layer (MPTMS).
Fig. 4. FT-IR spectra of the MPTMS layer (a) before and (b) after sintering at 780 °C.
Fig. 5. Variation of the membrane zeta potential with pH.
Fig. 3. Scanning electron microscopy of the UM01 membrane (a) and AUM01 membrane (b).
the TMP as expected from Darcy’s law. The hydraulic permeability deduced from the slope flux = f (TMP) indicated an almost three fold higher permeability for the UM01 membrane (88.4 L h−1 m−2 bar−1) than for the AUM01 membrane (31.7 L h−1 m−2 bar−1). A decrease in the flux was observed for both membranes during the first 30 min of operation before a steady-state was reached (Fig. 7). The same time was required to reach steady-state with salt and dye solutions. Lower flux obtained with salt solutions resulted from concentration polarization. Fig. 6. Water flux as a function of TMP.
4
Journal of Water Process Engineering xxx (xxxx) xxx–xxx
A. Chougui et al.
Fig. 7. Evolution of permeation flux through UM01 and AUM01 membranes over time for different feed solutions (TMP = 4 bar).
3.3. Retention rate of salts and dyes 3.3.1. Effect of concentration and ionic strength on salt rejection Salt rejection performance of both UM01 and AUM01 membranes obtained for single salt solutions (NaCl, MgCl2, Na2SO4 and MgSO4) in the pH range 5.8–6.2 and for a TMP of 4 bar is shown in Fig. 8. For both membranes and whatever, the electrolyte rejection rates were found to decrease with increasing feed concentrations (see Fig. 8(a) and (c)). This is the signature of a rejection mechanism ruled by the so-called Donnan exclusion. Salts with magnesium cations (MgCl2 and MgSO4) were less rejected than salts with sodium cations (NaCl, and Na2SO4), which is in line with negative zeta potentials measured for both membranes in the pH range 5.8–6.2 (see Fig. 5) since bivalent cations are more attracted by the negative membrane charge than monovalent cations. Fig. 8b and d confirm the dominant Donnan exclusion mechanism since a clear correlation between the salt rejection and the medium ionic strength was found. Indeed, an increase in the ionic strength leads to compression of the electrochemical double layer and then to weaker electrostatic repulsions. 3.3.2. Dye rejection Fig. 9 shows the variation of dye rejection as a function of concentration. The AUM01 membrane exhibits higher rejections than the UM01 membrane whatever the dye is. For the AUM01 membrane, rejection was found to be almost independent of the dye concentration (for the concentration range under investigation) while the UM01 membrane performance was more sensitive to dye concentration. This can be explained by the larger pores of the UM01 membrane as compared to the AUM01 membrane ones in addition to the weaker electrostatic interactions between charged dyes and the UM01 membrane. Interestingly, the rejection of the anionic dye, RB5, was lower than the one of cationic dyes, MG and NR although the pristine membranes were negatively charged (Fig. 5). A possible explanation is the adsorption of cationic dyes onto the negatively charged pores in the first times of filtration which causes a decrease in the pore size along the charge reversal surface. This leads to strong repulsive interactions between adsorbed and free MG and NR molecules. On the other hand, no (or at least limited) adsorption of RB5 is expected onto the like-charged membranes, which pore size is not reduced during RB5 filtration. 4. Conclusion
Fig. 8. Variation of salt rejection by UM01 and AUM01 membranes with salt concentration, (a) and (c), and ionic strength, (b) and (d).
The aim of this work was to develop, characterize and test new ultrafiltration ceramic membranes synthesized from a locally available barbotine, kaolin, zirconia and 3-Mercaptopropyltrimethyoxysilane. It was recognized from the tangential streaming potential measurements that membranes were negatively charged over a wide range of pH. Further, both salt and dye rejections were strongly influenced by
electrostatic interactions with the membrane surface. An anionic and two cationics were considered in ultrafiltration experiments. Very promising performance was obtained for the different dyes, with experimental rejections ranging from 80 to 100% for the 5
Journal of Water Process Engineering xxx (xxxx) xxx–xxx
A. Chougui et al.
References [1] A.J. Burggraaf, L. Cot, Fundamentals of inorganic membrane science and technology, Elsevier Science and Technology Series, No. 4, Elsevier, Amsterdam, 1996. [2] L. Xu, W.P. Li, S.Q. Lu, Z. Wang, Q.X. Zhu, Y. Ling, Treating dye waste water by ceramic membraneincrossflow microfiltration, Desalination 149 (2002) 199–203. [3] H. Zhang, X. Quan, S. Chen, H. Zhao, Y. Zhao, W. Li, Zirconia and titania composite membranesfor liquid phase separation: preparation and characterization, Desalination 190 (2006) 172–180. [4] J.-I. Oha, S.H. Lee, Influence of streaming potentialon flux decline of microfiltration with in-linerapid pre-coagulation process for drinking waterproduction, J. Membr. Sci. 254 (2005) 39–47. [5] L. Ricq, A. Pierre, J.-C. Reggiani, J. Pagetti, A. Foissy, Use of electrophoretic mobility and streamingpotential measurements to characterize electrokineticproperties of ultrafiltration and microfiltration membranes, Collids Surf. A 138 (1998) 291–299. [6] S. Condom, S. Chemlal, W. Chu, M. Persin, A. Larbot, Correlation between selectivity and surfacecharge in cobalt spinel ultrafiltration membrane, Sep. Purif. Technol. 25 (2001) 545–548. [7] A. Szymczyk, P. Fievet, M. Mullet, J.C. Reggiani, J. Pagetti, Comparison of two electrokineticmethods – electroosmosis and streaming potential — to determine the zeta-potential of plane ceramicmembranes, J. Membr. Sci. 143 (1998) 189–195. [8] Y. Elmarraki, M. Persin, J. Sarrazin, M. Cretin, A. Larbot, Filtration of electrolyte solutions withnew TiO2–ZnAl2O4 ultrafiltration membranes in relationwith the electric surface properties, Sep. Purif. Technol. 25 (2001) 493–499. [9] L. Ricq, A. Pierre, J.-C. Reggiani, J. Pagetti, Streaming potential and ion transmission during ultra-and microfiltration on inorganic membranes, Desalination 114 (1997) 101–109. [10] M. Nyström, M. Lindström, E. Matthiasson, Streaming potential as a tool in the characterization of ultrafiltration membranes, Collids Surf. A 36 (1989) 297–312. [11] P. Fievet, M. Sbaï, A. Szymczyk, C. Magnenet, C. Labbez, A. Vidonne, A new tangential streaming potential set-up for the electrokinetic characterization of tubular membranes, Sep. Sci. Technol. 39 (2004) 2931–2949. [12] A. Szymczyk, C. Labbez, P. Fievet, B. Aoubiza, C. Simon, Streaming potential through multilayer membranes, AIChE J. 47 (2001) 2349–2358. [13] P. Fievet, B. Aoubiza, A. Szymczyk, J. Pagetti, Membrane potential in charged porous membranes, J. Membr. Sci. 160 (1999) 267–275. [14] A. Szymczyk, P. Fievet, J.C. Reggiani, J. Pagetti, Characterisation of surface properties of ceramic membranes by streaming and membrane potentials, J. Membr. Sci. 146 (1998) 277–284. [15] A. Chougui, K. Zaiter, A. Belouatek, B. Asli, Heavy metals and color retention by a synthesized inorganic membrane, Arab. J. Chem. 7 (2014) 817–822. [16] A. Belouatek, N. Benderdouche, A. Addou, A. Ouagued, N. Bettahar, Preparation of inorganic supports for liquid waste treatment, Microporous Mesoporous Mater. 85 (2005) 163–168. [17] M. Smoluchowski, Zurtheorie der electrischenkataphorese und der oberfluchenleitung, Phys. Z 6 (1905) 529–536. [18] S.H. Christopher, B.M. Dennis, Some electrochemical and photoelectrochemical properties of 3-amino-7-dimethylamino-2-methylphenazine (neutral red) in aqueous solution, Aust. J. Chem. 36 (1983) 507–516. [19] S.K. Kim, B. Nordén, Methyl green: a DNA major-groove binding drug, FEBS Lett. 315 (1993) 61–64. [20] V. Hanykýř, J. Kutzendorfer, Technology of Ceramics, SilisPraha a Vega Hradec králové, Praha, 2000 (in Czech). [21] D.W. Breck, Zeolite Molecular Sieves, Wiley-Interscience, New York, 1974. [22] N. Hebalkar, S. Kharrazi, A. Ethiraj, J. Urban, R. Fink, S.K. Kulkarni, Structural and optical investigations of SiO2-CdS core-shell particles, J. Colloid Interface Sci. 278 (2004) 107–114. [23] B. Stuart, Modern Infrared Spectroscopy, John Wiley & Sons New York, NY, USA, 1996. [24] R. Takagi, M. Nakagaki, Characterization of the membrane charge of Al2O3 membranes, Sep. Purif. Technol. 25 (2001) 369–377. [25] K. Zaiter, A. Belouatek, A. Chougui, B. Asli, A. Szymczyk, Color, COD, and salt retention by inorganic membrane, Desalin. Water Treat. (2013) 1–7. [26] M. Kolaei, K. Dashtian, Z. Rafiee, M. Ghaedi, Ultrasonic-assisted magnetic solid hase extraction of morphine in urine samples by new imprinted polymer-supported on WCNT-Fe3O4-NPs: central composite design optimization, Ultrason. Sonochem. 33 (2016) 240–248. [27] M. Jamshidi, M. Ghaedi, K. Dashtian, S. Hajati, A. Bazrafshan, RSC Adv. 73 (2015) 59522–59532.
Fig. 9. Effects of dyes concentration on the retention.
most efficient membrane. Higher rejections were obtained with cationic dyes which were attributed to strong electrostatic repulsions between dyes and membranes after dye adsorption onto the membrane pores leading to (i) pore size reduction and (ii) surface charge reversal.
Acknowledgment We wish to thank gratefully Professor Anthony SZYMCZYK from the Institute of Chemical Sciences of Rennes who provided insight and expertise that greatly assisted the present manuscript.
6