Surface roughness is measured by Atomic Force Microscopy (AFM), which is also called scanning force microscopy (SFM). The AFM consists of a cantilever with a sharp tip (probe) at its end that is used to scan the specimen surface with a demonstrated resolution of an order of fractions of a nanometer. This device relies on electron tunneling in a narrow gap between the tip and the sample, where electrons jump from one side to the other through vacuum when two objects are close enough. Since the current is related to the distance of the tip and the sample, surface morphology can be mapped by scanning a sample surface.
The surface roughness is estimated by the arithmetic average of the absolute values of the surface height deviations measured from the center plane based on AFM images. Root mean square roughness is the standard deviation of the average roughness. The surface area difference represents the percentage increase of the three-dimensional surface area over the two-dimensional surface area (Hobbs, 2006).
It is physically straight forward to understand that rough surface is easier to attract particles than smooth surface. As illustrated in Fig. 1, particles can deposit to the zones with reverse flow due to the eddy occurring behind the bumps. The large contact area provided by the curved surface increases the chance of permanent particle settling by enhancing van deer Waals forces, charge interactions, chemical interactions, etc. This surface roughness may play a major role until the cake layer grows enough to make the initial surface roughness less significant.
Fig. 1. Particle deposition on rough membrane surface.
The effect of surface roughness, zeta-potential, and contact angle on percentage flux decline was investigated using 2 NF and 2 RO membranes with different surface characteristics (Vrijenhoek, 2000, 2001). As shown in Fig. 2, the relative flux decline by colloidal silica solution (200mg/L) of the four different membranes could be very well correlated with the surface roughness. In contrast, various extent of membrane fouling (1-J/J0) were observed although the four membranes had approximately same contact angles and surface zeta-potentials.
Similar observation was made when six different NF and RO membranes were compared in other study (Hobbs, 2000,2006). In this study, the ration of three dimensional areas to two dimensional areas was defined as surface area difference, where the rougher the surfaces are the higher the surface area differences are. It was found that the flux declines were very well correlated with the surface differences when contaminated ground water was filtered.
When yeast cell deposition rates were monitored using six different MF and UF membranes, the initial cell deposition rate was best correlated with surface roughness with a linear correlation factor of 0.79 (Kang, 2006). Zeta-potential was also positively correlated with the cell deposition rate and the linear correlation factor was 0.69. Although surface roughness and zeta-potential appear to have similar influences on membrane fouling, it is noteworthy that those correlation factors were obtained only for the single cake layer at most. In this experiment, particle deposition on a translucent polycarbonate flat sheet membrane was directly monitored by a optical microscope through the membrane from permeate side, which was the technique so called direct monitoring through membrane, DOTM (Li, 1998). Due to the limitation of the technique, only the monolayer formation could be observed. The membrane surface supposedly remained virgin without a significant coating by macromolecules until yeast cells deposited since the yeast particles were cleaned with DI water twice before the experiment. Although there was little hindrance in the charge interaction between the virgin membrane and the yeast cells, surface roughness was still found more significant in membrane fouling.
Fig. 2. Effect of surface roughness, contact angle, and zeta-potential on membrane fouling by 200mg/L 100 nm colloidal silica (or flux decline) at 51 LMH : LFC-1 (Hydranautics, Oceanside, CA), NF-70 (Dow Filmtec, Minneapolis, MN), X-20 (Trisep, Goleta, CA), HL (GE Osmomics, Minnetonka, MN), J = flux with colloidal silica, J0 = clean water flux (reproduced from Vrijenhoek, 2001).
© Seong Hoon Yoon