Theory – Streaming zeta potential
According to Stern’s model (Shaw, 1969), if charged surfaces are contacted with polyelectrolyte solution, the charge is balanced by counter ions (Fig. 1). The layer of the counter ion adsorbed on the surface is called Stern layer and the remainder of the counter ions are distributed in the diffuse layer. In case of negatively charged surface, the electric potential on the surface is the highest and it decreases rapidly until the Stern Plane due to the adsorbed positive ions. Outside the Stern Plane, the magnitude of potential decreases slowly.
Fig. 1. Electric double layer according to Stern’s model (after Shaw, 1969).
It is worthwhile to notice that the counter ion density near Stern Plane is affected by the flow velocity. As a result, the Stern potential is always affected by the experimental condition and cannot be measured accurately. Therefore, electro-kinetic zeta potential is measured instead, which represents the potential in shear plane of which thickness is affected by the hydrodynamic conditions of experiment.
The streaming zeta potential of surface is measured based on the voltage developed along the charged surface while electrolyte solution is flowing. Depending on the electric potential in Stern plane, the delay of counter ions varies and as a result varying degree of potential differences develop between upstream and downstream. Fig. 2 shows how streaming zeta potential is measured for membrane surface and pore wall.
Fig. 2. Streaming zeta potential of membrane surface (xS) and pore wall (xP). Voltage difference is measured upstream and downstream of the surface while electrolyte solution is flowing.
The streaming zeta potential of pore wall is measured while filtering electrolyte under pressure (Nyström, 1994). Likewise the streaming zeta potential of membrane surface is measured while flowing electrolyte over membrane surface (Elimelech, 1996). The relationship between the streaming potential and the zeta potential is given by the well-known Helmholtz-Smoluchowski equation (Elimelech, 1996).
ζ = zeta-potential (V)
ΔE = streaming potential (V)
ΔP = pressure difference across the channel (kg/m/s2)
μ = viscosity of the solution (kg/m/s)
εr = relative permittivity of the solution (-)
ε0 = electric permittivity of vacuum (F/m)
λ = conductivity of polyelectrolyte solution (S/m).
Impact of zeta potential on membrane fouling
Membranes with negative zeta-potential repulse negatively charged particles and macromolecules. When six different membranes with a range of zeta-potential between -5.5mV and -19.7mV were compared in terms of membrane fouling rate by the yeast cells (S. cerevisiae) with a zeta-potential of -8.7mV, a positive correlation was found. The linear correlation factor between the membrane zeta-potential and the yeast cell deposition rate was found at 0.69 (Kang, 2004). However, it is noteworthy that the experiment was done with pure yeast cells cleaned with DI water twice and the particle deposition rate was measured only in the initial phase of the surface coverage under a very low crossflow velocity (2.5cm/s). Therefore, yeast cell depositions were rarely affected by the yeast cells already deposited on membrane surface. In addition, the low back transport velocity under the exceptionally low crossflow velocity made charge repulsion the only meaningful back transport mechanism under the condition.
When PVDF membrane surface (Durapore® from Millipore, Watford, UK)) was modified with 2-acrylamido-2-methyl-1-propanesulfonic acid (AMPSA), 2-hydroxyethyl methacrylate (HEMA) and quaternized 2-(dimethylamino) ethyl methacrylate (qDMAEM), the modified membranes with hydrophilic surface were less susceptible to the fouling by E. coli than the membranes with hydrophobic surfaces. In the meantime, chemically neutral surfaces were easier to be cleaned than charged surfaces. The low cleaning efficiency was attributed to the charge attraction between membrane surface and E. colicells (Kochkodan, 2006). However, it is important to know the experiment was performed with pure E. coli washed three times with DI water and filtration was performed in a stirred cell at 300 rpm. The membrane cleaning was performed by running the stirred cell for 5 minutes with DI water without chemical cleanings. Therefore, the findings are not directly applicable for MBR, where various particles and biopolymers exist at high concentration and strong oxidants such as hypochlorite are used at high concentration (1,000-5,000 mg/L) as cleaning agents.
In MBR, various macromolecules with various chemical and physical properties exist in mixed liquor, which increases the chance of initial coatings of membrane as soon as mixed liquor contacts with membrane even before filtration starts. As explained with the figure here, some extent of initial membrane fouling is inevitable regardless of the flux in MBR. In addition, vigorous aeration is regularly performed, which makes 20-60 cm/s of upflow liquid velocity in a typical aeration condition (Yamanoi, 2010). Since membrane surface is not visible to particles in mixed liquor due to the initial deposit layer and membrane scouring is vigorous, charge repulsion between the original membrane surface and the particles is no longer an influential factor affecting particle deposition.
When four different membranes with different surface zeta-potentials (-40mV to -10mV) were tested to filter 1% whey protein, the surface zeta potential tended to converge to a similar level (~-30mV) at pH 8-10 regardless of the initial zeta-potential (Lawrence, 2006). In other experiment, the surface charge of a nanofiltration membrane that had a positive zeta-potential (+12 mV) initially turned to negative (-9mV) at neutral pH when the membrane was contacted with 18.5 mg/L humic acids solution (Yoon, 1998). The surface charge reversal was attributed to humic acid adsorption onto membrane surface. In other experiment, when membranes are negatively charged initially, they became more negative when they were exposed to humic acids (Elimelech, 1996). In other experiment, initial membrane surface property was hardly influential in further cake layer growth since solutes and particles in bulk liquid see only the cake layer . When 6 different RO and NF membranes were compared, no correlation between initial membrane zeta-potential and membrane fouling propensity was observed (Hobbs, 2000).
© Seong Hoon Yoon