Demhation, 95 (1994) 269-266 Eleevier Science B.V. Amsterdam - Printed in The Netherlands

269

Measuring the zeta (electrokinetic) potential of reverse osmosis membranes by a streaming potential analyzer

Menachem Elimelech,’ William H. Chen and John J. Waypa

Department of civil and Environmental Engineering, 4173 Engineering I, University of California, Los Angeles, CA !ZM24-1593 (USA)

(Received September 7, 1993; in revised form December 23, 1993)

SUMMARY

The use of a novel streaming potential analyzer to measure the zeta potential of cellulose acetate and composite polyamide reverse osmosis membranes is reported. Zeta potentials of these membranes were measured at various solution chemistries. These include effects of salt (NaCl) concentration, solution pH, and the presence of dissolved humic substances. It is demonstrated that streaming potential is a useful tool to measure zeta potential of reverse osmosis membrane surfaces. Results indicate that solution chemistry has a marked effect on the electrokinetic properties of reverse osmosis membranes. Humic substances strongly adsorb onto the surface of reverse osmosis membranes and thus alter the surface charge of the membranes. Furthermore, the zeta potential of reverse osmosis mem- branes becomes more negative as the NaCl concentration in solution increases, in a marked contrast to conventional electric double layer theories. It appears that the zeta potential of reverse osmosis membranes is strongly influenced by the presence of unreacted chemical substances or

*Author to whom correspondence should be addressed.

OOll-9164/94/$07.00 0 1994 Elsevier Science B.V. Au righte reserved.

SSDZOOll-9164(94)00064-U

270

impurities on the membrane surface. Various explanations for the behavior of the membranes at the above solution chemistries are evaluated and discussed.

Key words: Reverse osmosis membranes, zeta potential, membrane charge, streaming potential, electrokinetic charge, colloidal fouling

SYMBOLS

A -

Es - G - L - P - R - 4 -

Greek

& -

Eo - 9 -

!: - KS -

cross-sectional area of channel streaming potential free energy length of channel pressure electrical resistance electrical resistance of a standard solution

dielectric permittivity of water dielectric permittivity in vacuum absolute viscosity of water zeta potential specific conductivity of a standard solution

INTRODUCTION

Polymeric reverse osmosis (RO) membranes are used extensively in numerous separation processes, including sea water desalination and wastewater reclamation [l-3]. Fouling is a major problem in RO operation, resulting in product water flux decline and larger operating costs. Colloids and particulate matter are thought to cause the most fouling [4-71. The mechanisms of both in fouling RO membranes are poorly understood, though it is known that during fouling both colloids and particulates deposit onto the surface of RO membranes and thus increase the resistance to water flow through the membranes [5,6].

271

The interaction of colloidal particles with membrane surfaces in aqueous media is dependent on, among other variables, the zeta (electrokinetic) potentials of the membrane surface and suspended particles. These, in turn, are controlled by the surface chemistry of the membranes and colloidal particles, as well as by the chemistry of the solution [5,8]. Hence, the determination of zeta potential of RO membranes at various solution chemistries is of paramount importance.

When brought into contact with an aqueous electrolyte solution, poly- meric membrane surfaces acquire an electric surface charge through several mechanisms. These mechanisms include dissociation (ionization) of surface functional groups, adsorption of ions from solution, and adsorption of polyelectrolytes, ionic surfactants, and charged macromolecules [8]. The zeta potential is a measurable parameter, related to the charge and electric double layer of surfaces in aqueous solutions. For macroscopic surfaces, as in the case of polymeric RO membranes, the zeta potential can be determined by the streaming potential technique.

A limited number of studies have reported the use of streaming potential and electro-osmosis to characterize the electrokinetic potential of the interior pore surfaces in microporous membranes [g-12]. While the electrokinetic potentials of membrane pore surfaces can be important in colloidal fouling phenomena involving microfiltration or ultrafiltration membranes, these are probably not important in colloidal fouling of RO membranes. In RO membranes the pores (voids) are extremely small, on the order of a few angstroms [3]. It is most likely that, in colloidal fouling of RO membranes, colloids interact with the exterior surface of the membrane (i.e., the active layer), rather than with the interior pore surfaces.

Measurements of zeta potentials of RO membrane surfaces have not as yet been reported in the literature. The primary objective of this paper is to demonstrate the use of a novel streaming potential analyzer for measure- ment of zeta potential of RO membrane surfaces. The zeta potentials of cellulose acetate and composite polyamide membranes at various solution chemistries are reported. It is shown that solution chemistry plays an important role in controlling the zeta potential of polymeric RO mem- branes. Prior to presenting the experimental results, a discussion on the origin of surface charge of polymeric surfaces in aqueous solutions and a description of the basic principles and theory of the streaming potential method are presented.

272

ORIGIN OF SURFACE CHARGE OF POLYMERIC MEMBRANES

Membranes in the presence of water can have electrically charged surfaces. The surface charge characteristics of polymeric membranes are dependent on the chemical properties of the membrane and the chemistry of the solution. A discussion of the mechanisms by which membrane surfaces can acquire surface charge in aqueous solutions is provided below. These mechanisms include dissociation (ionization) of surface functional groups; adsorption of ions from solution; and adsorption of polyelectro- lytes, ionic surfactants, and charged macromolecules.

Dissociation of sur$ace functional groups

Several polymeric membranes contain ionizable surface functionalities, such as carboxylic (R-COO-), amine (R-NH:), and sulfonic (R-SO,) surface groups. Surface charge arises from the protolysis of these functional groups [ l&5,13-151. For example, the following dissociation reactions can take place:

R-COOH = R-COO- + H+ (1)

R-NH,+ = R-NH, + H+ (2)

R-SO,H = R-SO, + H+

The surface charge is dependent on the degree of ionization and, hence, the pH of the aqueous solution. As seen from the above reactions, at low pH values, a membrane surface with amine functional groups can be positively charged, while at moderate to high pH values, a membrane with carboxyl functional groups can be negatively charged. Jacobasch and Schurz [15] have presented expressions for determining the dissociation constants of functional groups on polymer surfaces from measured zeta potentials.

Adsorption of ions

Even in the absence of ionizable functional groups, polymeric membrane surfaces can acquire a surface charge through the adsorption of anions from solution [16-181. Preferential adsorption of anions has been suggested as a source of surface charge on non-ionogenic surfaces (i.e., surfaces with no ionizable functional groups), such as hydrocarbon droplets [16,17] and

273

hydrophobic polystyrene latex colloids [ 19-211. It has been postulated that anions can approach more closely to nonpolar or hydrophobic surfaces because they are less hydrated than cations. In this process a surface will acquire a negative electrokinetic potential due to the presence of anions beyond the plane of shear. Jacobasch and Schurz [15] have developed expressions for determining the adsorption free energy of ions onto polymeric surfaces by means of zeta potential measurements. In addition, they unequivocally demonstrated that, in the presence of 1:l simple electrolytes (such as NaCl), the zeta potential of various polymers becomes more negative as the contact angle of the polymer increases (i.e., as the polymer is more hydrophobic).

Ahorption of suflactants, polyelectrolytes, and charged macromolecules

Membrane surfaces can also acquire a surface charge through the adsorption of polyelectrolytes, ionic surf&ants, and charged macromole- cules [8,14,15]. The free energy of adsorption, AG,,, of a solute is given by 181

AGads = A%f - 4cd"

where AG,,, represents the surface affinity for the solute and AG,, is the solvent affinity for the solute. Thus, a solute that is hydrophobic in character will readily adsorb onto a solid surface. For polyelectrolytes adsorption arises from London-van der Waals interactions, hydrophobic bonding of nonpolar segments, hydrogen bonding, electrostatic attraction, and chemical reaction with surface functional groups. If the polyelectrolyte and the membrane surface are similarly charged, adsorption occurs if the non-electrostatic attraction is greater than the electrostatic repulsion. In this case adsorption is enhanced by increasing the ionic strength or by the presence of polyvalent counterions. If the polyelectrolyte and the surface are oppositely charged, then adsorption is dominated by electrostatic attraction.

THEORY AND PRINCIPLES OF STREATvlING POTENTIAL

Electrokinetic eflects

As described in the previous section, polymeric membranes acquire a

274

surface charge when brought into contact with an aqueous solution. The surface charge is compensated by counterions in the solution close to the surface, forming the so-called electrical double layer. The distribution of ions at the solid-liquid interface can be described by several models, which are discussed in detail elsewhere [22-241.

The vital feature of the electric double layer is that the surface charge is balanced by counter-ions, some of which are located very close to the surface, in the so-called Stern layer, with the remainder distributed away from the surface in the d@kse layer. An important parameter of the electric double layer is the Stern potential, that is, the potential at the boundary between the Stern and diffuse layers. The Stem potential cannot, however, be measured directly; the electrokinetic (zeta) potential is often considered as an adequate substitute. The zeta potential is the potential at the plane of shear between the surface and solution where there is relative motion between them. Several techniques are available that can be used to deter- mine the zeta potential of surfaces. Among these, the streaming potential technique is most suitable for membrane surfaces.

The relative motion between an electrolyte solution and a charged solid surface can result in one of four electrokinetic effects: (1) electrophoresis, (2) electro-osmosis, (3) sedimentation potential, or (4) streaming potential [22,23]. The induced electrokinetic effect depends on the driving force and the nature of the solid and liquid phases as schematically described in Fig. 1. Measurements made using techniques based upon each effect should, in principle, result in the same calculated value for the zeta potential. In practical situations, however, several factors conspire to produce results that are dissimilar. These include assumptions in the model for the electric double layer, inadequacies in the theory, and experimental errors related to the design and construction of the apparatus and sample preparation [22].

The zeta potential of flat surfaces, such as RO or nanofiltration mem- branes, can be measured by either the streaming potential or electro- osmosis method. The streaming potential method is preferred over electro- osmosis when measuring the zeta potential of flat surfaces because it is more convenient to measure small electrical potentials rather than small rates of liquid flow [22]. The principles and theory of streaming potential, as applied to this study, are described below.

275

Fig. 1. Illustration of the four electrokinetic effects and their principles.

Streaming potential

When an electrolyte solution is forced, by means of hydraulic pressure, to flow through a porous plug of material, across a channel formed by two plates, or down a capillary, a streaming potential is generated. The liquid in the channel carries a net charge. Its flow, due to hydraulic pressure, gives rise to a streaming current, thereby generating a potential difference (Fig. 2). This potential opposes the mechanical transfer of charge, causing back conduction by ion diffusion and electro-osmotic flow (due to the potential difference). The transfer of charges due to these two processes is called the leak current (Fig. 2). When equilibrium condition is attained, the streaming current cancels the leak current, and the measured potential difference is the streaming potential [22,23].

The relationship between the measurable streaming potential and the zeta potential is given [14,15]:

4 q L j-=--- 1 AP E.Q A i?

by the well known Helmholtz-Smoluchowski equation

(5)

276

a. Electrk Doable Layer at Rest

SOlOtiOn /

canterions

+ + + + + ++ +++ + m _

_

- s&f& -. _- -\ soIfaocclratge

b. Movement of Ions Due to Liquid Flow

Is Movmcnt of ions pmduces a s-g

c. Accumulation of Ions Dowmtream

E Accumulation causes + +

+ ++ formation of potmtial

+ + + *++++++ difference, E

-- _ _ _ - - _ _ -

d. Leak Curmnt Through Liquid

- ++++++ + Potential difference. E,

+ + +A+ + + causes leak current IL

--_ _ _* _ _ - -

Fig. 2. Schematic description of the various processes at the solid-liquid interface leading to the formation of streaming potential.

Here f is the zeta potential, E, is the induced streaming potential, AP is the applied hydraulic pressure, q is the liquid viscosity, E is the liquid permit- tivity, e. is the permittivity in vacuum, R is the electrical resistance across the medium, and L and A are the length and cross-sectional area of the channel, respectively.

The ratio L/A in Eqn. (5) consists of two parameters that cannot be determined easily for materials of complex geometry, such as granular media, cylinders, or porous membranes. Various approaches and models to determine this ratio have been proposed in the literature and discussed and analyzed by Jacobasch and coworkers [14,15]. For flat, nonporous surfaces such as RO membranes, the streaming potential is induced when a solution flows tangentially over the membrane surface. Hence, in this case L and A can be measured from the dimensions of the channel formed by the two pieces of membrane.

The value of each term in Eqn. (5) is either known or can be measured. The first term, EJAP, is determined by direct measurement of the stream-

277

ing potential generated for a given, constant, driving pressure. For water at a given temperature, the values of q and e are known. The electrical resistance across the channel, R, can be measured directly using an AC conductivity bridge. The L/A ratio can be determined as discussed above. Alternatively, using the Fairbrother and Mastin approach [15,25], this ratio can be replaced by K&, where R, is the electrical resistance of the channel when the measurement cell is filled with a standard solution whose specific conductance, K~, is known. A 0.1 M KC1 solution is commonly used to avoid problems associated with surface conductance.

Finally, it should be mentioned that the Helmholtz-Smoluchowski theory was originally derived for a single capillary, and Eqn. (5) is an extension for bundles of capillaries or for a channel formed by two plates. Several assumptions were made in the derivation of Eqn. (5), most of which have been critically discussed by Christoforou et al. [26], Hunter [23], and Cohen and Radke [27]. While some of these assumptions are questionable for granular and porous materials, no apparent limitations exist in applying Eqn. (5) to a channel formed by two flat surfaces.

EXPERIMENTAL METHODS

Streaming potential analyzer

The measurements presented in this paper were carried out using a novel streaming potential analyzer (BI-EKA, Brookhaven Instruments Corp., Holtsville, New York, USA). The instrument includes an analyzer, a data control system, and measuring cells. The analyzer consists of a mechanical drive unit to produce and measure the pressure that drives the electrolyte solution from a reservoir into and through the measuring cell. Circulation of the solution is achieved by a pump with a flow capacity of 1.3 l/min. Operation can be controlled manually or by computer. The streaming potential as well as the streaming current are measured simultaneously by the instrument. Sensors for measuring the temperature, pH, and conductivi- ty of the solution are also available.

There are two different measuring cells that can be used: cylindrical or rectangular. The choice of which cell to use depends on the nature of the sample under investigation. A cylindrical cell is used to measure the zeta potential of solid bodies such as fibers, granules, and rods. A rectangular

278

cell is suitable to measure the zeta potential of samples in the form of plates, films, or foils; this cell configuration is most suitable far RO membranes. The cell used in this research is made of PMMA and has dimensions of 125 ~50 mm. A channel of well defined dimensions (75 x 9.5 mm) is created by the use of PTFE spacers. The Ag/AgCl electrodes that measure the induced streaming potential are mounted at each end of the channel. Schematic illustrations of the electrokinetic analyzer and associated measuring cells were given by Jacobasch and Schurz [ 151,

Reverse osmosis membranes

Two types of membranes, cellulose acetate and composite polyamide, were used in this research (Hydranautics, San Diego, USA). The cellulose acetate membranes are made of a blend of cellulose diacetate and triacetate; no other details about their synthesis were given by the manufacturer. These membranes were supplied as wet, flat sheets and were stored in a 0.5 % formaldehyde solution. The interfacially polymerized thin film composite membranes were fully aromatic polyamide. They were supplied as dry flat sheets and were stored dry at room temperature. Details on the synthesis of the composite membranes and the various chemicals involved are given in [28].

Solution chemistries

Streaming potentials were measured as a mnction of pH and electrolyte concentration and in the presence of humic substances. The pH was adjusted by adding proper amounts of NC1 or KOH, while the electrolyte concentration was adjusted by adding NaCl. All chemicals used were analytical reagent grade (Fisher Scientific, Pittsburgh, USA). A stock solution of humic acid was made by dissolving commercial, dry humic acid (Aldrich Chemical Company, Milwaukee, USA) in deionized water adjusted to pH of about 11. The solution was filtered through a coarse glass paper filter followed by vacuum filtration with 0.22 pm membrane filters (Millipore Corp., Bedford, Massachusetts, USA). The TOC of the resulting stock solution was 1355 mg/L. The stock solution was stored in a refrigera- tor at 4°C.

279

Membrane preparation

Membranes were rinsed thoroughly and soaked in deionized water for 24 h prior to the streaming potential measurements. Following this step, they were cut into 125 X50 mm pieces to fit the rectangular cell. Two pieces of membranes were used for each measurement. One piece, with its active layer side facing down, was attached to the upper part of the cell; and the other piece, with its active layer side facing up, was attached to the lower part of the cell. The two pieces were separated by spacers that create a channel for the electrolyte solution to flow through. The cell was then tightened by screws on both ends and was attached to the streaming potential analyzer.

Determination of zeta potential

Before the start of each experiment, the cell was flushed with the electrolyte solution in both directions for a period of about 2 min to remove entrapped air bubbles from the cell. In a typical streaming potential measurement, the solution was pumped into the cell in alternating directions for a total of six times (three times in each direction). Each pumping period lasted 30 s. The reproducibility of the measurements was very good in both directions. The asymmetry potentials were below 0.5 mV, much smaller than the measured streaming potentials. The zeta potentials were calculated from the measured streaming potentials using Eqn. (5). The ratio L/A was determined using the Fairbrother and Mastin approach [ 15,251, as described earlier in this paper. The resistance of the channel, R, was calculated from the ratio of the streaming potential to the streaming current. Preliminary experiments demonstrated that, as expected, the generated streaming potential is linearly dependent on the applied pressure differential.

RESULTS AND DISCUSSION

Cellulose acetate membranes - Eflect of solution pH and ionic strength

The effect of solution pH and ionic strength on the zeta potential of the cellulose acetate membranes is shown in Fig. 3. The zeta potential of the membranes was measured over the pH range of 3-l 1 for two different

280

I - I

0. -o- 0.001 M NaCl .

s -10 - -a- 0.01 M NaCl .5

3 .w -2o- & 5 p. -30. rp % N-40-

I . I. I 2 4 6 8 10

PH Fig. 3. Zeta potential of the cellulose acetate membranes as a function of pH measured at hvo different NaCl concentrations.

NaCl solutions (0.001 and 0.01 M). Based on these results, the following observations are made: (1) the zeta potential is negative at all pH values investigated; (2) the zeta potential becomes more negative as the pH increases; and (3) for pH values greater than about 5, the zeta potential becomes more negative as the ionic strength increases.

Cellulose diacetate/triacetate membranes contain acetyl (CHsCO-) and hydroxyl (R-OH) functional groups as part of their polymeric structure. These functional groups, however, do not dissociate at the above chemical conditions. Hence, the negative electrokinetic charge acquired by the cellulose acetate membranes cannot be explained by dissociation of the above functional groups.

The negative surface charge of the cellulose acetate membranes can result from adsorption of anions (Cl- and OH’) from solution. This mechanism has been invoked to explain the negative zeta potential of a variety of non-ionogenic surfaces [16-181, as described earlier in this paper. The concentration of Cl- and OH- in solution increases with increasing NaCl concentration and pH and, as a result, the electrokinetic charge of the membranes becomes more negative. Jacobasch and coworkers [ 14,151 have demonstrated that various non-ionogenic polymer surfaces acquire a negative charge in aqueous electrolyte solutions. They also showed that the electrokinetic charge of polymer surfaces becomes more negative as the hydrophobic@ of the surface increases. This was explained by the fact that anions in aqueous solutions are less hydrated than cations

281

and thus approach hydrophobic surfaces more favorably than hydrophilic ones.

The usual acetylating agent for cellulose is acetic anhydride. Excess compound is rinsed from the polymer with water and rapidly hydrolyzes to acetic acid. Traces of this weak acid are probably adsorbed on polymer particles and may influence the surface charge. The zeta potential curves shown in Fig. 3 are characteristic of surfaces with weak acidic functional groups.

It should be noted that in the production of cellulose acetate RO mem- branes, various chemicals (e.g., organic swelling agents) are added to improve the performance of the membranes. While these chemicals are not part of the cellulose polymer structure, they can still be attached to the membrane surface as impurities and, thus, influence the electrokinetic charge. Unfortunately, no information could be obtained from the manufac- turer on the chemistry of these additives. It is possible that these impurities contain acidic functional groups, since the zeta potential of the membranes increases with pH, even at below neutral pH, much more than can be expected by adsorption of hydroxyl ions alone.

Composite membranes - Efect of solution pH and ionic strength

The effect of solution pH and ionic strength on the zeta potential of the composite polyamide membranes is shown in Fig. 4. As with the cellulose acetate, the zeta potential of the composite membranes was measured over the pH range of 3-11 for 0.001 and 0.01 M NaCl solutions. In addition, the zeta potential of the membrane was measured in deionized water over the above pH range. Based on the results shown iu Fig. 4, the following observations are made: (1) the membrane is positively charged at low pH and negatively charged at pH values above the isoelectric point (pH 3.5); (2) the zeta potential curves at all ionic strengths intersect at the isoelectric point, indicating that there is no specific adsorption of cations; and (3) the zeta potential becomes more negative as the pH and ionic strength increase.

Before discussing the above observations, a brief description of the processes and chemicals involved in the synthesis of the composite polyam- ide membranes is offered. (More details can be found in [28].) This description provides a better understanding of the surface chemistry of the composite membrane and thus can explain the zeta potential data shown in Fig. 4.

282

l”A-‘-‘-‘-’ -A- Deionized Water

-o- 0.001 M NaCl

-o- 0.01 M NaCi

I . I . , . , 2 4 6 6 10

PH

Fig. 4. Zeta potential of the composite polyamide membranes as a function of pH measured at three electrolyte solutions.

The composite membranes are prepared by interfacially polymerizing on a microporous support, a monomeric aromatic polyamine, such as pheny- lene diamine (in aqueous solution) with amine-reactive reactant comprising a polyfunctional acyl halide such as trimesoyl chloride (in organic solvent). It has been reported that the addition of monomeric amine salt to the aqueous solution above the microporous support can significantly enhance the performance of the membrane [28]. In addition to the amine salt, an anionic surfactant, such as sodium dodecyl benzyl sulfonate or sodium laurel sulfate, is added to the aqueous solution. Furthermore, in many cases acetic acid or other organic acids with carboxyl functional groups are added to the aqueous solution to adjust the PH. In the final stage the membranes are dried in an oven for a short period of time (less than 10 min).

It is hypothesized that small amounts of amine salts, unreacted mono- meric polyamines, or surfactants remain on the surface of the membrane after drying. Although these “impurities” are not part of the chemical structure of the aromatic polyamide, they can markedly influence the surface charge. In addition, hydrolysis of a third acyl chloride group of trimesoyl chloride to carboxyl can take place during interfacial polymeriza- tion, resulting in negatively charged carboxyl functional groups [29].

At low pH values, the amine salts and monomeric polyamines are positively charged. The anionic surfactants, on the other hand, are negative-

283

ly charged at low pH (above about 2), while the apparent carboxyl groups are uncharged. As a result, the overall zeta potential of the composite polyamide membranes is positive at pH values smaller than 3.5. As the pH increases, acidic functional groups of carboxyl and remaining surfactants deprotonate, thus impariing a negative charge to the membrane. As with the cellulose acetate, the zeta potential of the composite membranes is more negative at higher salt concentrations, most probably due to adsorption of Cl- from solution.

Efect of humic substances

Dissolved natural organic matter is ubiquitous in natural water bodies such as rivers, oceans, lakes, groundwaters, and estuaries [8,30,31]. The predominant fraction of dissolved natural organic matter in aquatic environ- ments comprises humic substances. Humic substances are refractory anionic polyelectrolytes of low to moderate molecular weight. They contain both aromatic and aliphatic components with mainly carboxylic and phenolic functional groups 18,321. Carboxylic functional groups account for 60-90% of all functional groups [33]. As a result, humic substances are negatively charged in the pH range of natural waters.

The configuration of humic substances is highly dependent on solution pH and ionic strength [8]. The characteristic lengths of humic substances are the hydrodynamic radius (when in solution) and the adsorbed layer thickness (when adsorbed on a surface). At high ionic strength or low pH, humic substances have a small hydrodynamic radius in solution and a large adsorbed layer thickness when adsorbed on the surface. On the other hand, at low ionic strength or high pH, humic substances have a large hydrody- namic radius and a small adsorbed layer thickness. Similar to polyelectro- lytes, humic substances are readily adsorbed onto solid surfaces in aqueous solutions and markedly affect their electrokinetic charge [8,34,35]. Since RO is used for desalination of various natural waters, it is imperative that the effect of humic substances on the zeta potential of these membranes be investigated.

The effect of humic substances on the zeta potential of the cellulose acetate and composite polyamide membranes is illustrated in Figs. 5 and 6, respectively. In the study with the cellulose acetate membranes, humic substances with TOC levels of 0.7 and 1.4 mg/L were used, while for the composite membranes TOC levels of 0.3 and 0.7 mg/L were used. The

O- 0 1V3 hi NaCl

-.- NoTOC

-O- 0.7 mg/LTOC - s l -A- 1.4mgLTOC . g

-3

Fig. 5. Zeta potential of the cellulose acetate membranes as a function of pH measured in the presence of humic sub- stances. The concentration of humic sub- stances is expressed as TOC (see figure legend). All experiments were carried out in the presence of 0.001 M NaCl as a background electrolyte.

1W3 M NaCl -A- NoTOC

-O- 0.3 mglL TOC _ -e- 0.7mgRTOC .

I. I. 1 - 1 * a I 2 4 6 6 -10

PH Fig. 6. Zeta potential of the composite polyamide membranes as a function of pH measured in the presence of humic sub- stances. The concentration of humic sub- stances is expressed as TOC (see figure legend). All experiments were carried out in the presence of 0.001 M NaCl as a background electrolyte.

background NaCl concentration in these experiments was low3 M. As observed in Figs. 5 and 6, humic substances have a marked effect on the zeta potential of RO membranes. For the composite membranes, the zeta potential in the presence of humic substances is negative over the entire pH range of 3-l 1. Furthermore, for both membranes, the zeta potential is more negative in the presence of humic substances.

Humic substances affect the zeta potential of membranes through adsorption onto the membrane surface. The negatively charged functional groups of humic substances dominate the surface charge of the RO mem- branes. It is now recognized that humic substances control the zeta potential of surfaces and particles in aquatic environments [34-361. In natural waters, most particles and surfaces are negatively charged due to adsorption of hu- mic substances. The TOC levels used in this study are typical of many surface waters and sea waters which may be used as feed waters for RO and nanofiltration.

CONCLUSIONS

The use of a novel streaming potential analyzer to measure the zeta

285

potential of RO membranes was reported. Effects of solution pH, ionic strength, and the presence of humic substances on the zeta potential of cellulose acetate and composite polyamide membranes were investigated. Based on the results reported in this paper, the following conclusions are made:

1. Streaming potential can be effectively used to measure the zeta potential of RO membranes.

2. Cellulose acetate and composite polyamide membranes are negatively charged at all pH values of practical interest (pH larger than 3.5).

3. The zeta potential of both membranes becomes more negative as the pH and the ionic strength of the solution increase.

4. Cellulose acetate and composite polyamide membranes acquire a negative surface charge through adsorption of anions from solution.

5. Humic substances adsorb onto membrane surfaces and markedly affect their zeta potential.

6. Chemical substances on membrane surfaces, which were introduced during the synthesis of the membranes, can markedly influence the electro- kinetic charge and zeta potential of polymeric membranes.

Future research in this area should focus on measuring the zeta potential of pure cellulose acetate and aromatic polyamide polymers, rather than using commercial membranes whose surface properties are not completely known. Such research can result in a better understanding of the mecha- nisms involved in the development of surface charge of polymeric mem- branes in aqueous solutions. It is also suggested that the effect of various cleaning techniques on the zeta potential of RO membranes be investigated. Cleaning can remove adsorbed impurities from the membrane surface and thus alter the surface charge.

ACKNOWLEDGMENTS

The research reported in this paper was funded by the State of Califor- nia, Department of Water Resources (DWR), and by the National Water Research Institute (NWRI). We would like to thank Brian Smith and Kurt Kovac from DWR and Ronald Linski from NWRI for their support and en- couragement.

286

REFERENCES

1 M. Mulder, Basic Principles of Membrane Technology, Kluwer Academic, Dordrecht, 1991.

2 3

H.K. Lonsdale, J. Membrane Sci., 10 (1982) 81. R.E. Kesting, Synthetic Polymeric Membranes: A Structural Perspective, 2nd ed., John Wiley, New York, 1985. D.E. Potts, R.C. Ahlert and S.S. Wang, Desalination, 36 (1981) 235. R.D. Cohen and R.F. Probstein, J. Colloid Interface Sci., 114 (1986) 194. G. Belfort, G. Alexandrowicz and B. Marx, Desalination, 19 (1976) 127. G. Belfort, Desalination, 34 (1980) 159. W. Stumm, Chemistry of the Solid-Water Interface, Wiley Interscience, New York, 1992. M. Tasaka, S. Tamura, N. Takemura and K. Morimoto, J. Membrane Sci., 12 (1982) 169.

10 M.G.A. Khedr, S.M. Abdel Haleem and A. Baralm, J. Electroanal. Chem., 184 (1985) 161.

11 12 13

14 15 16 17

18 19 20 21 22 23

24 25 26

27 28

J.A. Ibanez, J. Forte, A. Hemandez and F. Tejerina, J. Membrane Sci., 36 (1988) 45. M. Nystrom, M. Lindstriim and E. Matthiasson, Colloids Surfaces, 36 (1989) 297. N.V. Churaev, I.P. Sergeeva, V.D. Sobolev and D.E. Ulberg, J. Colloid Interface Sci., 151 (1992) 490. H.-J. Jacobasch, G. Baubijck and J. Schurz, Colloid and Polymer Sci., 263 (1985) 3. H.-J. Jacobasch and I. Schurz, Progr. Colloid and Polymer Sci., 77 (1988) 40. D.O. Jordan and A.J. Taylor, Trans. Faraday Sot., 48 (1952) 346. H.A. Abramson, Electrokinetic Phenomena and Their Application to Biology and Medicine, ACS Monograph Ser. No. 66, Chem. Catalog Co., New York, 1934. I. Stachurski and M. Michalek, Colloids Surfaces, 15 (1985) 255. L.P. Voeglti and C.F. Zukoski, J. Colloid Interface Sci., 141 (1991) 92. M. Elimelech and C.R. O’Melia, Colloids Surfaces, 44 (1990) 165. J.R. Goff and P. Luner, J. Colloid Interface Sci., 99 (1984) 469. D.J. Shaw, Electrophoresis, Academic Press, London, 1969. R.J. Hunter, Zeta Potential in Colloid Science: Principles and Applications, Academic Press, London, 1981. J. Westall and H. Hohl, Adv. Colloid Interface Sci., 12 (1980) 265. F. Fairbrother and H. Mastin, J. Chem. Sot., 75 (1924) 2318. CC. Christoforou, G.B. Westermann-Clark and J.L. Anderson, J. Colloid Interface Sci., 106 (1985) 1. R. Cohen and C. Radke, J. Colloid Interface Sci., 141 (1991) 338. E. Tomaschke, Interfacially Synthesized Reverse Osmosis Membrane Containing an Amine Salt and Processes for Preparing the Same, United States Patent 4,948,507, August 14, 1990.

29

30 31

J.E. Cadotte, in: D.R. Lloyd (ed.), Materials Science of Synthetic Membranes, American Chemical Society, Washington, D.C., 1985. W. Stumm and J.J. Morgan, Aquatic Chemistry, Wiley Intemcience, 1981. F.M.M. Morel and J.G. Hering, Principles and Applications of Aquatic Chemistry, Wiley Interscience, New York, 1993.

32 K. Gosh and M. Shnitzer, Soil Sci., 129 (1980) 266. 33 E.M. Perdue, Geochim. Cosmochim. Acta, 42 (1978) 1351. 34 K.A. Hunter and P.S. Liss, Nature, 282 (1979) 823. 35 L. Liang and J.J. Morgan, Aquatic Sciences, 52 (1990) 32. 36 CR. O’Melia, Colloids Surfaces, 39 (1989) 255.