Ceramic membranes are gaining more and more attention due to their inherent advan- tages compared with polymeric membranes. Their high thermal, mechanical and chem- ical stability make them more applicable in treating e.g. corrosive and hot wastewater. With their small pore sizes, the ceramic tight ultrafiltration (tight UF) membranes show a greater selectivity than ceramic UF or microfiltration (MF) membranes. However, low permeability and unstable membrane quality were observed using most commercially available ceramic tight UF membranes.
Atomic layer deposition (ALD), have come up as an important technique for deposit- ing thin films, and have turned out to be a good potential alternative to produce tight UF ceramic membranes from ceramic UF or MF membranes, due to its capability of con- trolled deposition of a one single atom layer film.
In this study, we explored the potential of ALD in the fabrication of tight UF ceramic membranes from normal sol-gel made UF ceramic membranes with various initial pore sizes. Since the resistance against flux within a porous membrane is mainly from the separation layer, we primarily focused on the coating depth of the coated membranes by the ALD technology. In addition a relation was found between the coating depth and the performance of the coated membranes.
The results of the experiments demonstrate that ALD method could successfully be used in the fabrication of tight UF ceramic membranes and, to some extent, cure the defects in the original membranes. By applying different ALD settings, different coating depths in the membrane filtration layer were achieved, which influenced the water per- meability and solute rejection. Further, membranes with larger pore sizes tended to have a deeper effective coating depth since the precursors could diffuse more easily through the larger pores. The Carmen Kozeny model, used to determine the permeability and porosity for each studied membrane, showed results in the same order of magnitude with most of the measurement data, being thus able to predict permeability as a func- tion of porosity and layer thickness.