You are on page 1of 7

Colloids and Surfaces A: Physicochemical and Engineering Aspects 138 (1998) 283289

Physicochemical aspects of polymer selection for ultraltration and microltration membranes


E.R. Cornelissen, Th. van den Boomgaard *, H. Strathmann
Department of Chemical Engineering, University of Twente, P.O. Box 217, NL-7500 AE Enschede, Netherlands Received 17 June 1996; accepted 1 September 1996

Abstract The concept of additivity of surface tension components has been used to predict the adsorptive fouling tendency of membranes. The calculated value for the free energy of adhesion DG is taken as a measure for this fouling LWS tendency. DG values can be determined from the surface tension components of the solid and liquid. The LWS determination of the surface tension components of the solid surfaces has been carried out using sessile drop contact angle measurements with water, glycerol and a-bromonaphthalene. DG values have been determined for human serum albumin (HSA) and polyethyleneglycol (PEG) on ten LWS dierent non-porous polymer surfaces and it has been calculated that cellulose acetate (CA) and polyacrylonitrile (PAN ) showed the lowest fouling tendency, whereas polyethersulfone (PES) and polyvinylidenediuoride (PVDF ) showed the highest fouling tendencies. 1998 Elsevier Science B.V. Keywords: Adsorptive fouling; Membranes; Polymers; Surface tension

1. Introduction One of the major problems encountered during ultraltration and microltration is the occurrence of a decline in ux over time due to concentration polarization and fouling. Fouling of membranes is a complex phenomenon usually the result of adsorption of feed solution components onto the membrane material. The general picture in membrane technology is that membranes prepared from hydrophilic materials are better than hydrophobic materials with respect to adsorptive fouling. However, the use of hydrophilic materials is not always possible since the chemical as well as the mechanical stability of most hydrophilic materials
* Corresponding author. Fax: +32 53 4894811; e-mail: A.vandenBoomgaard@ct.utwente.nl 0927-7757/98/$19.00 1998 Elsevier Science B.V. All rights reserved. PII S0 9 2 7 -7 7 5 7 ( 9 6 ) 0 3 86 2 - 9

is signicantly lower than that of hydrophobic materials. The selection of a proper membrane material for a certain feed solution is a matter of optimization between chemical stability of the membrane material and fouling tendency of the feed solution components. The prediction of the fouling tendency is at the moment still largely empirical. The aim of this paper is to predict the fouling behavior of polymer membranes by feed solution constituents applying the general physicochemical concept of additivity of intermolecular forces at interfaces. This concept, developed by Fowkes [1] and then by van Oss [2,3] to interpret contact angle measurements, can be used to characterize surfaces [4,5] and to calculate the adsorptive fouling tendency. Contact angle measurements are dicult to interpret on surfaces with pores. To a

284

E.R. Cornelissen et al. / Colloids Surfaces A: Physicochem. Eng. Aspects 138 (1998) 283289

rst approximation, we assume that the eect of pores on the fouling tendency of membranes can be neglected. Therefore, at surfaces of membrane polymers are used.

2. Theory 2.1. Interaction forces in condensed media The existence of the solid and liquid states is due to the short-range forces of attraction between molecules in these states. The phenomenon of surface or interfacial tension arises from these forces, because of a net attraction of molecules from the surface toward the bulk. Surface and interfacial tension are thus dened as the normal forces on any line at the surface or more fundamentally as the work that is required to reversibly increase the surface area at constant temperature, pressure and composition [6 ]: c=

ments can be carried out with a high-energy apolar liquid. In condensed media, polar or Lewis acidbase (AB) interactions can be divided into electrondonor and electron-acceptor parts and are essentially asymmetric. The polar contribution of the surface tension of a certain material thus consists of two parts, the electron-donor parameter part cD and the electron-acceptor parameter part cC. The total polar component of the surface tension cAB can be found from the geometric mean between the two parameters: cAB=2 cCcD (2)

A B
G

A T,P,n

(1)

According to the IUPAC nomenclature [6 ], surface tension is dened at the boundary of two phases one of which is a gas, while interfacial tension is dened at the boundary of two liquids or at a solidliquid interface. The short-range intermolecular forces, responsible for surface and inter-facial tensions, are usually divided into dierent classes. According to van Oss [2,3], four dierent classes can be distinguished. Usually, these dierent interaction forces do not inuence each other. The dierent interaction forces or surface tension components are: (1) apolar or Lifshitzvan der Waals (LW ) forces (cLW); (2) polar or Lewis acidbase (AB) forces (cAB); (3) electrostatic ( EL) forces (cEL); (4) interaction forces due to Brownian movement (BR) (cBR). To determine the apolar contribution of the surface tension of a certain solid material (e.g. a at, hard polymer surface), contact angle measure-

In order to determine the two parts of the polar surface tension component cAB experimentally, contact angle measurements have to be carried out with two dierent polar liquids. If a material consists of charged particles or molecules, electrostatic forces have to be taken into account. Electrostatic interactions are determined with the help of electro-kinetic measurements, such as streaming potential measurements and electrophoresis. In our case electrostatic interaction forces will play a minor role and will be neglected. van Oss introduced a xed contribution for the interaction forces due to Brownian motion, resulting from the thermal energy of molecules. This contribution is usually very small compared to the other three interaction forces and will be neglected in our calculations. The dierent interaction forces are assumed to be additive. The total surface tension is thus given by the sum of the two important contributions: c=cLW+cAB=cLW+2 cCcD (3)

In the next section a method is discussed to obtain these surface components. 2.2. Contact angle measurements Contact angle measurements can be used to determine the surface tension of an unknown surface. These measurements can be carried out by the sessile drop technique [79] and the captive bubble technique [8,10,11]. These two techniques

E.R. Cornelissen et al. / Colloids Surfaces A: Physicochem. Eng. Aspects 138 (1998) 283289

285

Dupre equation: DG =c c c (5) LS LS L S in which DG is the work or the free energy of LS adhesion. The free energy of adhesion of a liquid and a solid immersed into a third liquid w, usually water, is given by: =c c c (6) LWS LS LW SW By substitution of Eq. (3), as was found by van Oss, the following equation can be derived: DG DG LWS =2[ cLW cLW + cLW cLW cLW cLW cLW L W S W L S W (7)

Fig. 1. Dierent techniques to determine contact angles.

are depicted in Fig. 1 from which their basic dierences become clear. In the captive bubble technique the surface is already in the wet state, unlike the sessile drop technique. This is an advantage because the surface does not have to be dried. The sessile drop technique is, however, a faster technique which easily gives a contact angle. In performing contact angle measurements with the sessile drop technique, a droplet is placed upon a at homogeneous surface and the contact angle of the droplet with the surface is measured. The shape of the droplet depends on the balance between the solidliquid interfacial tension (c ), LS the liquidgas surface tension (c ) and the solidL gas surface tension (c ). This balance is given by S the Young equation and is indicated in Fig. 2. At equilibrium this balance is given by: c =c +c cos h (4) S LS L where h is a nite angle of contact between the liquid droplet and the solid surface. 2.3. The free energy of adhesion When a liquid which is not in contact with the solid substrate makes contact and adheres to it, the liquidgas interface area decreases. The work that is necessary for this adhesion is given by the

+ cC ( cD + cD cD )+ cD ( cC W L S W W L + cC cC ) cC cD cD cC ] S W L S L S The tendency of interaction between the liquid and the solid immersed in a certain medium (e.g. water) is thus expressed in DG . If the value of LWS the free energy of adhesion exceeds zero, no interaction occurs between the solid and the liquid. If this value is negative, adhesion will take place. This criterion can be taken as a tool for adhesion of fouling agents onto non-porous surfaces, and can also play an important role in the phenomena of membrane fouling. We only need to know the values of the surface and interfacial tension components. The determination of these components will be discussed in the next paragraph. 2.4. Determination of the surface tension components By combining Eq. (4) and Eq. (5), YoungDupre equation is obtained: the

DG =c (1+cos h) (8) LS L If we substitute Eq. (3) and Eq. (5) in this equation, we obtain the YoungDupre equation in terms of the components of the surface and interfacial tension as discussed by van Oss: (1+cos h)c =2( cLW cLW + cC cD + cD cC ) L S L S L S L
Fig. 2. The Young equation.

(9)

286

E.R. Cornelissen et al. / Colloids Surfaces A: Physicochem. Eng. Aspects 138 (1998) 283289

This equation holds for materials which do not carry electrical charges and are built up of polar and apolar interaction forces only. The three components of the surface tension of an unknown material can be found by performing contact angle measurements with three materials of known surface tensions. The apolar contribution of the surface tension, cLW of the solid surface, S can be determined by contact angle measurements with a high-energy apolar liquid. The two parts of the polar surface tension component cAB can be S determined by contact angle measurements with two dierent polar liquids. A set of three equations with three unknowns is formed, which can easily be solved. 2.5. Contact angle hysteresis In performing contact angle measurements, a smooth and homogeneous surface is required. In the literature contact angle measurements have also been carried out on rough surfaces [12] (e.g. membrane surfaces [8,10,11,13,14]), but the Young equation has then to be modied. For these surfaces a so-called contact angle hysteresis exists between the advancing and the receding contact angle, unlike smooth surfaces [7]. The advancing contact angle is measured when a new phase advances over the surface, whereas the receding contact angle is found when a phase retreats. Both

contact angles can be measured by the sessile drop and captive bubble techniques as shown in Fig. 3. The captive bubble technique will usually rst give a receding angle while the sessile drop method rst gives an advancing angle. When an advancing contact angle is measured, the surface encounters a new phase, which might lead to surface rearrangements, e.g. orientation of hydrophilic groups in polymer chains to the surface when in contact with water [8,12]. This might also be the origin of the contact angle hysteresis.

3. Experimental 3.1. Materials Polymer solutions were prepared from dierent polymers and solvents. Polymer materials were selected on the basis of their use as ultraltration and micro-ltration membrane materials in membrane ltration. The polymers used were cellulose acetate (CA) from BDH Chemicals; polyacrylonitrile (PAN ) from Aldrich Chemical Co. Inc.; polycarbonate (PC, 0.4 mm membranes) from Nuclepore; polyetherimide (PEI, Ultem 1000) from General Electric; polyethersulfone (PES, Victrex 4800) from ICI; polypropylene (PP, Carlon) from Shell; polysulfone (PSf, Udel P1800) from Amoco; polytetrauoroethylene (PTFE, thread seal tape type R10 D1) from Giveg and polyvinylidenediuoride (PVDF, Solef 2008) from Solvay. In Table 1 an overview is given of the dierent polymer solutions. The solvents used were
Table 1 The composition of the polymer solutions used for the preparation of polymer lms Polymer material CA PAN PC PEI PES PP PSf PTFE PVDF Amount (wt.%) 5 15 15 15 15 Pressed 15 Tape 15 Solvent Acetone Dimethylformamide Chloroform Chloroform N-methylpyrrolidone Not relevant Chloroform Not relevant Dimethylformamide

Fig. 3. Advancing (h ) and receding (h ) contact angles for the a r sessile drop and the captive bubble techniques.

E.R. Cornelissen et al. / Colloids Surfaces A: Physicochem. Eng. Aspects 138 (1998) 283289

287

acetone (Merck pro analysi>99.5%), dimethylformamide (DMF, Merck pro synthese >99%), chloroform (Merck pro analysi, 9999.4%) and Nmethylpyrrolidone (NMP, Acros>99%). The polymer solutions were cast onto glass plates with a casting knife of 0.15 mm and dried in a nitrogen atmosphere for about 6 h to prevent absorption of water from air. The PES in NMP lm was dried overnight in a vacuum oven at 80C. Contact angle measurements were carried out on the freshly prepared dry lms. Ultrapure water, glycerol (Merck pro analysi>99.5%) and a-bromonaphthalene (Merck pro synthese>98%) were used to determine respectively the two polar components cC , cD and the S S apolar component cLW of the surface tension of S the polymer surface. The known surface tension components of the three liquids are listed in Table 2. Ultrapure water was obtained from a MilliPore Q+ installation, in which demineralized water was fed and puried. The electrical resistivity of the ultrapure water was 18.2 MV cm at 25C. The water was organic-free with a total organic carbon content ( TOC ) of <10 ppb and is ltered through a 0.22 mm microltration membrane. 3.2. Methods Contact angle measurements were carried out on the dried lms with a Kruss G1 contact angle meter, with the three dierent liquids according to the concept of van Oss. Contact angles were measured by a direct method, that is by reading the angle between the tangent at the contact point and
Table 2 Surface tension components of model liquids and fouling agents [13] Material cLW (mJ m2) 21.8 44.4 34.0 26.8 43.0 cC (mJ m2) 25.5 0 3.92 6.3 0 cD (mJ m2) 25.5 0 57.4 50.6 64.0 c (mJ m2) 72.8 44.4 64.0 62.5 43.0

the horizontal line of the solid surface, using a protractor. Only advancing contact angles have been measured, since all the investigated surfaces were expected to be smooth and homogeneous. This would result in a negligible contact angle hysteresis. The tabulated values in Table 3 are the average of twelve measurements: three liquid droplets (two angles per droplet) on two dierent polymer lm samples; they were measured directly after applying the droplet onto the surface (within about 1 min).

4. Results and discussion A change in contact angles with time because of surface rearrangements due to the change of surrounding phase (air was replaced by a liquid), as was reported in the literature [8,11], has not been observed because of the short measuring times. According to others [7,15] an equilibrium contact angle has been found within 5 min for porous membranes. Short measuring times for contact angle measurements were used to prevent evaporation eects of the liquids. From Table 3 the polar and apolar components of the interfacial tension can be calculated from the theory of van Oss. These values are listed in Table 4 for the dierent surfaces. The free energy of adhesion which can be calculated from Eq. (8), will be used as a measure for the fouling tendency. To calculate this energy for fouling substances on polymer surfaces in water, the three surface tension components of every
Table 3 Average values of the measured contact angles (in degree) of water, glycerol and a-bromonaphthalene Polymers CA PAN PC PEI PES PP PSf PTFE PVDF h water 593 573 781 792 922 942 822 1172 922 h glycerol 543 494 662 632 685 833 674 1122 1043 h a-bromonaphthalene 262 61 121 81 132 421 147 932 292

Water a-Bromonaphthalene Glycerol HSA PEG

288

E.R. Cornelissen et al. / Colloids Surfaces A: Physicochem. Eng. Aspects 138 (1998) 283289

Table 4 Calculated interfacial tension components of the dierent polymers Polymers cLW S (mJ m2) 40 44 44 44 43 34 43 10 40 cC S (mJ m2) 0.5 0.6 0.1 0.3 0.5 0 0.2 0 0 cD S (mJ m2) 19 19 5.8 3.9 0.1 1.7 3.1 0.9 0.1 c (mJ m2) 46.2 50.8 45.5 46.2 43.4 34.0 44.6 10.0 40.0

CA PAN PC PEI PES PP PSf PTFE PVDF

fouling compounds. This is contrary to PES and PVDF which show the lowest free energy of adhesion for HSA and PEG in water. From the sign of DG in Fig. 4, it can be seen LWS that HSA only adheres to PES and PVDF; for all the other polymer materials, DG for HSA in LWS water is positive, so on the basis of this concept adsorptive HSA fouling can only be expected for these two materials. For PEG no adhesion onto CA and PAN can be expected, because of the positive sign of DG . All other materials display a negative free LWS energy of adhesion, indicating a certain tendency to adhesion.

substance (polymer, fouling agent and water) have to be known. From the literature [3], human serum albumin (HSA) and polyethyleneglycol (PEG) were taken as standard fouling agents, because of their wide use as model fouling agents in membrane ltration [1621]. The three surface tension components are tabulated in Table 2. The calculated free energies of adhesion between the dierent polymers and the two fouling agent materials are plotted in Fig. 4. From this gure it can be seen that among all investigated polymers CA and PAN display the highest free energy of adhesion for HSA and PEG in water. These materials are expected to have the lowest degree of irreversible fouling during membrane ltration for HSA and PEG. From the literature [22,23] it is known that these materials indeed show a low fouling tendency for other

5. Conclusion The concept of additivity of intermolecular forces at interfaces can be used to calculate the free energy of adhesion of model fouling agents onto polymer lms, which gives an indication of the adsorptive fouling tendency of ultraltration and microltration membranes. It has been seen that CA and PAN polymers show the highest free energy of adhesion. These materials are also known in membrane technology as low fouling membrane materials. On the other hand, PES and PVDF showed the lowest energy of adhesion indicating a high adsorptive fouling tendency. As was mentioned before, the discussed concept neglects the eects of pore size and pore size distributions which do have a large eect on membrane fouling in practice. Apart from this shortcoming, the concept of van Oss is a powerful tool in predicting the adsorptive fouling tendency of ultraltration and microltration membranes.

Acknowledgment One of the authors ( E.R.C.) thanks the Unilever Research Laboratory Vlaardingen for nancial support. Thanks are due to Patrick Wuis who carried out the contact angle measurements and to Sander Rekveld for many helpful discussions about the van Oss theory.

Fig. 4. Calculated free energies of adhesion of HSA and PEG.

E.R. Cornelissen et al. / Colloids Surfaces A: Physicochem. Eng. Aspects 138 (1998) 283289

289

References
[1] F.M. Fowkes, in K.L. Mittal ( Ed.), Physico-Chemical Aspects of Polymer Surfaces, Vol. 2, Plenum, New York, 1983, p. 583. [2] C.J. van Oss, Acidbase interfacial interactions in aqueous media, Colloids Surfaces A: Physicochem. Eng. Aspects 78 (1993) 149. [3] C.J. van Oss, Interfacial Forces in Aqueous Media, 1st edn., Marcel Dekker, New York, 1994, p. 440. [4] T.B. Lloyd, K.E. Ferretti, J. Lagow, A new approach to chemical characterization of polymer surfaces, J. Appl. Polym. Sci. 58 (1995) 291296. [5] L. Gourley, et al., Characterization of adsorptive fouling on ultraltration membranes by peptide mixtures using contact angle measurements, J. Membrane Sci., 97 (1994) 283289. [6 ] D.H. Everett, Manual of symbols and terminology for physicochemical quantities and units, Pure Appl. Chem. 31 (1972) 577638. [7] F.F. Zha, A.G. Fane, C.J.D. Fell, R.W. Schoeld, Critical displacement pressure of a supported liquid membrane, J. Membrane Sci. 75 (1992) 6980. [8] W. Zhang, M. Wahlgren, B. Sivik, Membrane characterization by the captive angle technique. II. Characterization of UF-membranes and comparison between the captive bubble and sessile drop as methods to obtain water contact angles, Desalination 72 (1989) 263273. [9] M. Oldani, G. Schock, Characterization of UF membranes by infrared spectroscopy, ESCA and contact angle measurements, J. Membrane Sci. 43 (1989) 243258. [10] V. Gekas, K.M. Persson, M. Wahlgren, B. Sivik, Contact angles of ultraltration membranes and their possible correlation to membrane performance, J. Membrane Sci. 72 (1992) 293302. [11] W. Zhang, B. Hallstrom, Membrane characterization using the captive bubble technique. I. Methodology of the captive bubble technique, Desalination 79 (1990) 112.

[12] M. Morra, E. Occhiello, F. Garbassi, Knowledge about polymer surfaces from contact angle measurements, J. Colloid Interface Sci. 173 (1990) 1124. [13] K.J. Kim, A.G. Fane, C.J.D. Fell, The eect of LangmuirBlodgett layer pretreatment on the performance of ultraltration membranes, J. Membrane Sci. 43 (2-3) (1989) 187204. [14] A.G. Fane, C.J.D. Fell, K.J. Kim, The eect of surfactant pretreatment on the ultraltration of proteins, Desalination 53 (1985) 3755. [15] H. Takeuchi, M. Nakano, Progressive wetting of supported liquid membranes by aqueous solutions, J. Membrane Sci. 42 (1989) 183188. [16 ] S. Sridhar, P.K. Bhattacharya, Limiting ux phenomena in ultraltration of kraft black liquor, J. Membrane Sci. 57 (1991) 187206. [17] C. Bhattacharjee, P.K. Bhattacharya, Flux decline analysis in ultraltration of kraft black liquor, J. Membrane Sci. 82 (1993) 114. [18] M. Lopez-Leiva and E. Matthiasson, Solute adsorption as a source of fouling in ultraltration, in Fundamentals and Applications of Surface Phenomena Associated with Fouling and Cleaning in Food Processing, 1981. Tylosand. [19] E. Matthiasson, The role of macromolecular adsorption in fouling of ultraltration membranes, J. Membrane Sci. 16 (1983) 2336. [20] H. Reihanian, C.R. Robertson, A.S. Michaels, Mechanisms of polarization and fouling of ultraltration membranes by proteins, J. Membrane Sci. 16 (1983) 237258. [21] H.D.W. Roessink, et al., Characterization of new membrane materials by means of fouling experiments. Adsorption of BSA on polyetherimidepolyvinylpyrrolidone membranes, Colloid Surfaces, 55 (1991) 231243. [22] A.G. Fane, C.J.D. Fell, A review of fouling and fouling control in ultraltration, Desalination 62 (1987) 117136. [23] J.H. Hanemaaijer et al., Characterization of clean and fouled ultraltration membranes, Desalination, 68(2-3) (1988) 93108

You might also like