Separation and Purification Technology 44 (2005) 235–241

Tubular microporous alumina structure for demulsifying vegetable oil/water emulsions and concentrating macromolecular suspensions S´ergio R. Fontes a, ∗ , Viviane M. Silva Queiroz a , Elson Longo b , Marcus V. Antunes a a

S˜ao Carlos School of Engineering, University of S˜ao Paulo – USP, Department of Mechanical Engineering, Av. Trabalhador S˜ao-carlense, 400, C.P. 359, 13566-590 S˜ao Carlos, S˜ao Paulo, Brazil b Department of Chemistry, Federal University of S˜ ao Carlos – UFSCar, S˜ao Carlos, SP, Brazil Received 1 December 2004; received in revised form 1 February 2005; accepted 1 February 2005

Abstract A microstructure composed of alumina–silica (mullite, 3Al2 O3 ·2SiO2 ) was molded into tubes to be used in a microfiltration process for separating water/vegetable oil emulsions and to concentrate macromolecular suspensions. The microporous tubes were produced by the precipitation method using raw material supplied by Rhodia do Brasil Ltda, and sintered at a final temperature of 1450 ◦ C. The microporous medium was characterized by mercury porosimetry and by scanning electron microscopy (SEM). Most of the samples showed average pore sizes ranging from 0.3 to 20 ␮m, which the literature indicates as appropriate for the demulsification of macroemulsions. The microfiltering performance of the tubes was evaluated using emulsified mixtures of water and vegetable oil (sunflower and soybean) and macromolecular mixtures of xanthan and guar gum suspensions (molecular weight of 106 Da), under transmembrane pressures of 1.5–5.0 bar and a turbulent crossflow regime (Re > 10.000). The process was then repeated and the tubes’ performance compared with that of a commercial membrane of German origin with a nominal pore size of 0.4 ␮m. The quality of the permeate, from the standpoint of carbon retention in the mixtures, was evaluated based on measurements of the total organic carbon (TOC) and the pH. In the case of mixtures in suspension, the microporous tubes exhibited better carbon retention than the membrane. In the case of the emulsified mixtures, carbon retention exceeded 90%, and the demulsifying process achieved results compatible with those reported in the literature. © 2005 Elsevier B.V. All rights reserved. Keywords: Porous tubes; Alumina; Microfiltration; Emulsions; Suspensions

1. Introduction Porous alumina ceramics are commonly used as filters, refractory sensors, and as base material for the fabrication and coating of membranes [1,2]. Microfiltration and ultrafiltration membranes are usually made of aluminum oxide (Al2 O3 ) doped with zirconium or titanium [3,4] and produced by sintering. Membranes of this type are used in the separation of emulsive mixtures [3,5].

∗

Corresponding author. Tel.: +55 16 3373 9531; fax: +55 16 3373 9402. E-mail address: [email protected] (S.R. Fontes).

1383-5866/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.seppur.2005.02.001

The process of microfiltration using membranes is viable in several industrial applications, as in the concentration of suspensions or juices, in the production of pure liquids, and in the regeneration of emulsified mixtures [6,7]. Microfiltration, in particular, is used in the production of emulsions to control the globule size of the oily liquid phase, due to superficial tension phenomena acting on the permeable surface [8]. Growing industrial activity and the increasing generation of wastes by the petrochemical, metallurgical and food industries, as well as the high cost of treated water, make it imperative for studies to search for new materials to support separation processes. Studies of the mechanical properties of the tubular microstructure [1] have demonstrated its potential

236

S.R. Fontes et al. / Separation and Purification Technology 44 (2005) 235–241

use as an alternative to membranes owing to its easy, low cost production. In the study reported on here, a mixture of mullite (3Al2 O3 ·2SiO2 ) was used as an ingredient in the fabrication of a tubular microstructure for microfiltering emulsions and suspensions. The experimental results presented here physically characterize the microporous medium and its performance in the process for emulsions of water–vegetable oil (sunflower and soybean) and macromolecular suspensions of xanthan and guar gum. The results of transmembrane flow are presented as a function of processing time and physical properties (carbon concentration and pH) of the permeate.

2. Experimental investigation Eight 600-mm-long tubes with 10-mm internal diameter and 2-mm wall thickness were prepared by the sintering method, using mullite (3Al2 O3 ·2SiO2 ) supplied by Rhodia do Brasil Ltda. A 10 ◦ C/min heating rate was used to heat the electric furnace to the final sintering temperature of 1450 ◦ C. The microporous tubular structures were processed according to the most common ceramic processing technique, called the precipitation or slip casting method [1,9]. In this work, the “casting” was done at room temperature, while the “slip” was a powder–water mixture that was poured into a tubular mold. Much of the water was absorbed by the mold, leaving a relatively rigid powder body that was removed from the mold. To develop a strong tubular product, the piece had to be heated. Firing was done at higher temperatures, typically above 1000 ◦ C, which is a sensitive parameter in the sintering of tubular microporous bodies. This processing produced an asymmetric microporous structure. The tubes were subjected to the microfiltration process on an experimental stainless steel bench, as illustrated in Fig. 1. The experimental system was built with a positive displacement pump, manufactured by Netzsch do Brasil Ltda to produce transmembrane pressures of up to 10 bar. This pressure was kept constant by a frequency inverter mounted in the pump.

Fig. 1. Schematic drawing of the experimental apparatus: (1) jacketed fluid container; (2) pump; (3) flowmeter; (4) covering module with tubular microstructure; (5) outflow of the permeate.

In the microfiltration experiments, 2 vol.% concentrations of water/soybean or water/sunflower oil emulsions were used after intense shaking in a blender for 2 min. The oil was the disperse phase in this biphasic mixture. Macromolecular polyssacharides solutions were also produced with xanthan and guar gums [10] in six different proportions in a mass concentration of 1000 ppm. The polyssacharide composite solutions (guar and xanthan gums) were of high molecular weights (guar gum, 106 Da and xanthan gum, 4 × 106 Da), which are values reported in the literature with variations of up to one order of magnitude [10]. For this mixture, the use of the membrane with a nominal 0.4 ␮m cut-off is compatible with the microfiltration of the solutions’ high molecular weight, i.e., greater than 106 Da [7]. The soybean and sunflower oil dispersed phases were analyzed in an optical microscope (LEICA, model DMLB with software Imeid Pro Plus 4.1). A digital photograph of a sample of the initial water/soybean oil macro-emulsion concentrate is shown in Fig. 2. The distribution of the droplets revealed a significantly variable size ranging from 1 to 25 ␮m. The water/sunflower oil macro-emulsion displayed a similar pattern of droplets. In this work, emulsions were not produced for specific droplet sizes, so the influence of droplet size on the demulsifying process was not studied. The demulsifying process occurs according to the physical interaction of each phase of the biphasic mixture (macroemulsion) with the inorganic surface. Water has an affinity for mass transfer at microporous surfaces; hence, it is hydrophilic to permeable surfaces. However, the dispersed phase (oil) is hydrophobic to permeable surfaces. Fig. 3 illustrates the mass transfer at the permeable surface of each phase of the biphasic mixture confined in the microporous tube and subjected to a transmembrane pressure (P) of over 1 atm. Fig. 3b shows a water/oil droplet interacting superficially with the permeable surface. The water phase (hydrophilic) has an affinity for transferring through the permeable surface due to interfacial and superficial stresses,

Fig. 2. Digital photograph of a sample of initial water/soybean oil macroemulsion concentrate taken with an optical microscope.

S.R. Fontes et al. / Separation and Purification Technology 44 (2005) 235–241

237

Fig. 3. Microfiltration demulsifying process.

which differ substantially from the physical properties of the oily phase (hydrophobic). In the macromolecular concentration process, exclusion of the suspended phase (guar and xantham gums) prevails, according to the membrane’s nominal cut-off (nominal porosity) and the positive influence of the polarization layer [7]. In each case, the process was studied for at least 60 min and conducted so as to cause the mixture’s concentration to increase over time. A cooling system kept the temperature of the mixture stable at 25 ± 0.5 ◦ C. The transmembrane pressure was kept within an interval of 1.5–5.0 bar by a marking gauge installed at the end of the module. The permeate outflow through the microporous tube was regenerated by passing a neutral detergent and the NaOH base through the tube for at least 30 min between experiments. The permeate collected from both suspensions and emulsions maintained the characteristics of transparency of clean water. The mass transmembrane flux or permeate outflow was measured in all the experiments using a precision balance of two significant digits. The mass transmembrane flux was transformed into volumetric transmembrane flux based on the measurements of the permeate’s density using a density meter at the temperature of the permeate. The permeate’s physicochemical properties were investigated based on measurements of the total organic carbon (TOC) and the pH measured, respectively, with a Shimadzu (TOC 5000A) and an Orion model 290A pH meter, after a standard equipment calibration procedure. Measuring the transmembrane flux and using proper instrumentation were procedures that ensured the optimal accuracy of the experimental results of this work, with an experimental error of less than 10%.

3. Results and discussion 3.1. Characterization of the microporous medium Two tubes were selected, which were representative of the average pore size of the microporous medium. These tubes were characterized by the mercury intrusion technique (Autopore II 9220, Micrometrics Instruments Corporation) and by scanning electron microscopy (SEM-EDS-Stereoscan 440). Specimens selected from the tip of each tube were subjected to porosimetric and structural analysis under an electronic microscope. Fig. 4a and b illustrate the results of the pore size characterization of two tubes selected for experimentation (tubes 1 and 2). Fig. 4 shows the rate of variation in specific mercury volume (Hg) as a function of pore diameter. The typical distribution pattern [11] indicated that the predominant pore size of tube 1 (Fig. 4a) ranged from 0.3 to 10 ␮m, while the mean pore size of tube 2 was 0.3 ␮m. Tube 2 displayed an atypical behavior unlike the others, which showed characteristics similar to those illustrated in Fig. 4a (tube 1), with pore sizes varying at the most from 0.3 to 20 ␮m. The analysis of another three samples of tubes 1 and 2 indicated the same pore sizes as those depicted in Fig. 4. This analysis, which was carried out on samples of the ends and center of the microporous tube after it was used, indicated the good homogeneity of the microporous structure in terms of average pore size. Fig. 5 shows the internal surface of tube 1 analyzed structurally by SEM. Part of the sample used for porosimetry testing (Fig. 4a) was selected for this analysis. As can be seen, the morphological structure of the microporous

238

S.R. Fontes et al. / Separation and Purification Technology 44 (2005) 235–241

Fig. 4. Mercury extrusion porosimetry. Rate of variation of specific volume as a function of tube diameter: (a) tube 1 and (b) tube 2.

Fig. 5. SEM micrograph of the sample from tube 1: (a) magnified by 7.500× and (b) contrasts indicating differentiated composition.

medium is composed of grains with an average size of 1–4 ␮m. Fig. 5b shows a view of the same sample, contrasting the chemical composition of the microporous medium. The portions were selected in the figure from a white region, identified by the digit “1” and from a gray region (predominant) identified by the number “2”. The chemical composition of each region was characterized by SEM and EDS (energy dispersive spectroscopy). In region “1” (Fig. 5b), the medium’s localized composition was characterized by the presence of zirconium (58.3%), aluminum (35.0%), magnesium (2.4%) and oxygen (4.3%). Region “2” (Fig. 5b), representing most of the microstructure’s composition, showed a predominance of aluminum (75.9%) and minor amounts of magnesium (2.3%), oxygen (6.1%) and e silicon (0.4%). A comparison of these tubes with images of commercial membranes also revealed a strong morphological similarity [2].

of the water/soybean oil and the water/sunflower oil mixtures. The process was monitored for 1 h, with samples of the permeate collected at 5 min intervals. In all the tests, as indicated in Fig. 6, the transmembrane outflow at pressures of 1.5, 3.0 and 5.0 bar declined steadily right from the beginning of the process, characterizing the formation of a polarization layer on the tube’s porous surface [7,13]. The lowest transmembrane outflow values were achieved with 1.5 bar pressure for water/soybean oil and with 3 bar pressure for water/sunflower oil. The highest values

3.2. Characterization of the microporous medium in the microfiltration process Fig. 6 shows the results of transmembrane outflow (J in l/h m2 ) as a function of time (h) achieved in the microfiltration of oily emulsions (soybean and sunflower). The full and empty points in the figure represent, respectively, the results

Fig. 6. Transmembrane flux (l h/m2 ) as a function of time (min) for emulsions: (a) water/soybean oil, tube 1 and (b) water/sunflower oil, tube 2.

S.R. Fontes et al. / Separation and Purification Technology 44 (2005) 235–241

239

Fig. 7. pH as a function of time taken from samples of permeates from the microfiltration of water/soybean oil (a) and water/sunflower oil (b) emulsions.

of “J” correspond to 3 and 5 bar, respectively, for the water/soybean oil and water/sunflower oil. The results of “J” corresponding to 3 bar (water/sunflower oil) were very similar to those obtained in the same process with 1.5 bar. In cases such as this, the lack of major variations in the “J” outflow was attributed to the detergent’s insufficient cleansing action during cleaning of the microporous medium. Systematic reductions in transmembrane outflow resulting from the phenomenon of fouling-type incrustation are also common [12]. Fig. 6 depicts the results reported in the literature [13,14] for the sake of comparison with those of this study. The results of Nabi et al. [13] were obtained with ultrafiltration ceramic membranes for olive oil emulsions added to a surfactant (sodium dodecyl sulfate). As can be seen in the figure, the transmembrane outflow value obtained under a pressure of 1.5 bar was lower than the values achieved in this study. Arnot et al. [14] tested a zirconium membrane with a mean pore size of 0.1 ␮m, using a hydrocarbide emulsion (dode-

canal). They obtained a higher transmembrane outflow value under conditions inferior to those of this study (0.8 bar), and values close to those achieved in our study under pressures of 3.0 or 5.0 bar. Although there are differences in the membranes used and in the nature of the emulsions, this comparison is valid to check whether the transmembrane outflow values achieved in our study reached the same order of magnitude as those achieved with membranes. Fig. 7 shows the values of pH as a function of time for the same samples analyzed in Fig. 6 for transmembrane outflow. The pH value exceeded 7 (basic pH) in each case, with a tendency to reach 7.5 at the end of the process. Fig. 8 depicts the values of total organic carbon (TOC), measured with a Shimadzu (TOC 5000A) instrument as a function of time, of some of the permeate samples analyzed during the process (Fig. 6). Most of the TOC values obtained from the microfiltration of water/soybean oil emulsions (Fig. 8a) were lower than 10 mg “C”/l except those obtained under a pressure of 5 bar, which reached a TOC value of 40 mg “C”/l. In the latter case, we believe the pres-

Fig. 8. Total carbon—TOC (mg/l) as a function of time (min) of the permeate resulting from the microfiltration of water/soybean oil (a) and water/sunflower (b) emulsions.

240

S.R. Fontes et al. / Separation and Purification Technology 44 (2005) 235–241

sure applied in the process caused oil to pass through the microporous medium. The TOC value of the concentrated biphasic mixture (water/soybean oil macro-emulsion) was 2662 mg “C”/l. The TOC value of the concentrated biphasic mixture (water/sunflower oil macro-emulsion) was 3162 mg “C”/l. Thus, the two processes provided a significant reduction in the order of 98%, with most of the samples of permeate showing TOC values very similar to those of tap water (8–10 mg “C”/l). The permeate collected in this study displayed the characteristics of density, transparency and viscosity of clear water, indicating that the microporous medium separated the aqueous phase from the biphasic mixture, leaving the permeate in the minimal condition of reusable water for ordinary washing purposes. Fig. 9 depicts the experimental results of transmembrane outflow (J, l/h m2 ) as a function of time for the macromolecular suspensions combining xanthan and guar. All the experiments revealed a declining transmembrane outflow characterized by the formation of a polarization layer peculiar to the microfiltration process [7]. The experimental results shown in Fig. 9a correspond to the process undertaken with the microporous tube (tube 2). As can be seen in Fig. 9a and b, the process involving exclusively xanthan suspensions (1000 ppm) showed high transmembrane outflow values, while the aqueous mixtures with guar gum (1000 ppm) showed lower permeate outflow values. The experiments illustrated in Fig. 9b were reproduced using a commercial membrane of German origin with a nominal pore size of 0.4 ␮m (Fig. 9b). It is important to note that typical outflow behavior as a function of time was similar in the two processes, i.e., with the microporous tube (Fig. 9a) and with the commercial tubular membrane (Fig. 9b). The initial transmembrane outflow provided by the membrane was about twofold higher than that of the microporous tube. However, these values were approximately the same by the end of the process.

Fig. 9. Transmembrane outflow with different concentrations of guar and xanthan: (a) microporous tube and (b) commercial membrane (0.4 ␮m).

Fig. 10 shows the total organic carbon (TOC) of samples of permeate and of the initial concentrate (Fig. 9). The even numbers represent the results of the permeate samples while the odd numbers correspond to the initial mixtures. The sequential paired bars in the figure represent the mixtures indicated in the legend of Fig. 9b. An analysis of the results indicates that, in each case, the use of the microporous tube provided a total organic carbon

Fig. 10. Initial measurements of the concentrate and mean filtrate values of total organic carbon (TOC in mg “C”/l) for the aqueous guar/xanthan solutions and pure aqueous solutions: (a) microporous tube (tube 2) and (b) commercial membrane (0.4 ␮m).

S.R. Fontes et al. / Separation and Purification Technology 44 (2005) 235–241

retention (TOC) of over 90% (Fig. 9a), while the commercial membrane provided a lower organic carbon retention of 79–93% (Fig. 9b). Hence, the microporous tube provided a better concentration performance of combined xanthan and guar suspensions.

4. Conclusions The following conclusions can be drawn from the results of this study: 1. The microporous tubes performed well in the process with emulsions, showing transmembrane outflow values very similar to those obtained with membranes. 2. In the microfiltration of emulsions, the resulting permeate was a clear, transparent aqueous phase with basic pH of about 7.5, and the microporous tube reduced the TOC values of the initial mixture by at least 90%. 3. The best operating conditions were obtained at transmembrane pressures of 1.5 and 3.0 bar. At a pressure of 5 bar, the permeate’s TOC values were higher, probably because this pressure favored the intrusion of the oily phase. 4. The TOC and transmembrane outflow values obtained in the tests with macromolecular suspensions indicated that the microporous tube had a similar performance in comparation to the commercial membrane with a nominal pore size of 0.4 ␮m.

Acknowledgements The financial support of FAPESP (S˜ao Paulo State Research Support Foundation, Brazil) is gratefully acknowledged. Special thanks are also due to Engineer Luis Fernando Porto (Director of Tecnicer Ltda, S˜ao Carlos, S˜ao

241

Paulo, Brazil) for producing the microporous tubes used in this study. References [1] Z. Deng, T. Fukasawa, M. Ando, Microstructure, Mechanical properties of porous alumina ceramics fabricated by the decomposition of aluminum hydroxide, J. Am. Ceram. Soc. 84 (2001) 11. [2] K. Hwang, H. Lee, S. Lee, K. Chung, S. Park, J. Lee, Synthesis of aluminium hydrates by a precipitation method and their use in coatings for ceramic membranes, J. Eur. Ceram. Soc. 21 (2001) 375–380. [3] S.H. Hyun, G.T. Kim, Synthesis of ceramic microfiltration membranes for oil/water separation, Sep. Sci. Technol. 32 (1997) 2927–2943. [4] J. Sekulic, et al., Microporous silica and doped silica membrane for alcohol dehydration by pervaporation, Desalination 148 (2002) 19–23. [5] T.V.R. Alicieo, et al., Membrane ultrafiltration of crude soybean oil, Desalination 148 (2002) 99–102. [6] J.K. Ting, J.C.S. Wu, Cross-flow ultrafiltration of oil/water emulsions using porous ceramic membranes, J. Chin. Inst. Chem. Eng. 30 (1999) 207–214. [7] L.J. Zeman, A.L. Zydney, Microfiltration and Ultrafiltration— Principles and Applications, Marcel Dekker, New York, 1996. [8] S.M. Joscelyne, G. Tr¨agardh, Membrane emulsification—a literature review, J. Membr. Sci. 169 (2000) 107–117. [9] J.F. Shackelford, Introduction to Materials Science for Engineers, fourth ed., Prentice-Hall, New Jersey, 1996. [10] R. Lapasin, S. Pricl, Rheological of Industrial Polyssacharides: Theory and Applications, Blackie Academic and Professional, Glasgow, 1995. [11] S.J. Gregg, K.S.W. Sing, Adsorption, Surface Area and Porosity, second ed., Academic Press, London, 1982. [12] S. Ripperger, J. Altmann, Crossflow microfiltration—state of the art, Sep. Purif. Technol. 26 (2002) 19. [13] N. Nabi, P. Aimar, M. Meireles, Ultrafiltration of an olive oil emulsion stabilized by an anionic surfactant, J. Membr. Sci. 166 (2000) 177–188. [14] T.C. Arnot, R.W. Field, A.B. Koltuniewicz, Cross-flow and deadend microfiltration of oily-water emulsions. Part II. Mechanisms and modeling of flux decline, J. Membr. Sci. 169 (2000) 1–15.