Efficiency of chemical free cleaning for fouling control in membrane processes: influence of fouling properties
Dependence on membrane technology in seawater desalination and wastewater reclamation is increasing rapidly due to potable water shortages across the world. Nanofiltration (NF) and reverse osmosis (RO) are widely used membrane filtration techniques which use hydraulic pressure as the driving force for mass transport through the membrane. Interest is growing in use of forward osmosis (FO) membrane technology in wastewater reclamation and as a pre-step to low pressure reverse osmosis due to potential energy reductions. In FO, two liquids with different osmotic pressures are separated by a membrane and the osmotic pressure difference is the driving force for water permeation from the feed side to the draw side, eliminating the need for hydraulic pressure. Membrane fouling is an unavoidable problem facing all membrane processes. The presence of contaminants such as organic waste and bacteria in the feed solution can lead to membrane fouling and to a reduction in performance. This study aims to research cleaning methods that don’t involve the use of harmful chemicals for fouling control in FO and RO membrane processes. Cleaning efficiency is tested under varying physical and chemical conditions. Efficiency is determined in terms of both flux restoration and fouling layer removal as it has been determined in this study that significant flux restoration does not translate to complete fouling layer removal. Feed solution chemistry, such as the quantity of Ca2+ in the feed and the operating pressure, and therefore initial flux greatly influence the rate and extent of fouling and therefore the efficiency of cleaning. Understanding fouling behaviour during operation and cleaning is important when optimising cleaning methods for RO and FO. For this reason, this study uses numerous visualisation and surface characterisation techniques to improve understanding of the fouling structure before and after cleaning. The first aim of this study is to examine the efficiency of osmotic backwashing as a way of controlling organic fouling of BW30 membranes through both pure water flux measurements and membrane surface imaging. Confocal laser scanning microscopy is used to determine the thickness of the fouling layer before and after osmotic backwashing. Firstly, the influence of feed solution chemistry was examined. After 6.5 hours of organic fouling with alginic acid, the fouling layer thickness increased from 37 μm in the absence of calcium in the fouling solution to 179 μm in the presence of 2.5 mM CaCl2. This occurred due to the formation of thick compact gel fouling layer as carboxyl groups present in alginate fouling form complexes with Ca2+. One minute of backwashing with 0.7 M NaCl resulted in initial pure water flux restorations of 90% for the membrane fouled in the absence of CaCl2 compared to 86% for the membrane fouled in the presence 2.5 mM CaCl2 despite the fact that 31 μm and 141 μm of foulant remained on these membranes respectively. As well as the feed solution chemistry, the impact of initial flux, on fouling and cleaning in RO was examined. As expected, higher initial fluxes resulted in thicker fouling layers with the layer increasing from 39 μm for an initial flux of 25 L/m2h to 270 μm for an initial flux of 100 L/m2h. Although flux restorations of 100% were achieved for initial fluxes of 25 and 33 L/m2h, foulant layers of 9 μm and 25 μm remained on the membrane surface respectively. This again shows that flux restoration alone cannot indicate cleaning efficiency as high values have been reported even when significant fouling remains on the membrane surface. Only 75% of the initial pure water flux was restored after fouling with an initial flux of 100 L/m2h showing that organic fouling at higher initial fluxes is largely irreversible due to the high hydraulic pressure applied to the fouling layer making it more dense and compact. As the solution chemistry of the feed solution has a significant influence on the fouling layer characteristics, it is questioned whether the backwashing solution could alter the fouling layer characteristics during one minute of backwashing. In order to examine this, backwashing with a solution of 0.5 M CaCl2 was tested. SEM-EDS measurements showed that the amount of elemental calcium on the membrane decreased by 19% after backwashing with NaCl but increased by 26% after backwashing with the same osmotic pressure of CaCl2. Atomic force microscopy was used to quantify the adhesion forces and elasticity of the fouling layers. The membrane backwashed with CaCl2 displayed adhesion forces twice that of the virgin membrane due to the presence of Ca2+ ions forming complexes with carboxyl groups in the fouling layer. In terms of the elastic forces, the sample backwashed with NaCl displayed forces similar to that of a virgin membrane showing that the fouling layer is stiff and firm. This is because the layer becomes much thinner and closer to the membrane surface. The fouled membrane was the most flexible and “fluffy” however after backwashing with CaCl2 the fouling layer remains flexible and soft, again showing that Ca2+ forms complexes with the fouling layer. SEM was also used to compare the morphology of the fouling layer before and after backwashing. Different backwashing trends are observed in osmotic backwashing of organically fouled forward osmosis membranes. In this case, the alginate fouling layer thickness increased from less than 33 μm, in the presence of 0 mM Ca2+ in the feed to 173 μm in the presence of 2.5 mM Ca2+. One minute of backwashing removed almost 100% of the fouling layer in each case and restored 93% of the flux in the absence of calcium and 100% for the membrane fouled in the presence of 2.5 mM CaCl2. Backwashing became less effective as the initial membrane fouling flux was increased. This was increased by increasing the draw solution concentration. For a draw solution of 4 M NaCl the fouling layer decreased in thickness to 113 μm but backwashing with 0.7 M NaCl only removed 19% of the fouling layer showing that although it was thinner, the fouling layer was, in fact, more compact and dense. In this case, despite 90 μm of fouling remaining on the membrane surface, 100% of the flux was restored. As with the case of organic fouling with RO, this result shows that flux restoration alone cannot indicate how effective membrane cleaning is. To test the true efficiency of backwashing in organic fouling of FO membranes, 5 consecutive fouling and backwashing cycles were performed. Even after 5 cycles, backwashing with 0.7 M NaCl can still remove the fouling layer and restore the initial pure water flux to 97%. This shows that backwashing is effective for the FO membranes subjected to organic fouling. The limitations of backwashing in FO are evident in the context of initial bacteria adhesion on FO membranes. Initial adhesion of Pseudomonas putida on forward osmosis membranes was performed for 30 minutes resulting in an 18% membrane surface coverage of live bacteria cells. Backwashing with 0.7 M NaCl was ineffective while backwashing with 3 M NaCl offered a higher backwashing flux and therefore removed 93% of the adhered cells. After 60 minutes of bioadhesion under the same conditions, 3 M NaCl backwashing was not effective as 13% of the membrane surface remained covered in dead cells. In order to test how the feed solution chemistry effects cell adhesion, varying concentrations of CaCl2 were added to the feed resulting in an increase in live cell surface coverage from 18% in the absence of calcium to 27% in the presence of 5 mM CaCl2. This is due to a reduction in the cell-surface separation distance as the Ca2+ ions reduce the repulsive force between the cells and membrane surface therefore increasing adhesion. This increase in adhesion resulted in a decrease in backwashing efficiency as the 3 M NaCl backwashing solution only removed 40% of the total bacteria. Backwashing with solutions of CaCl2 was tested to determine if 1 minute of backwashing was enough time for Ca2+ ions to influence the cells on the membrane surface. As with organic fouling, the Ca2+ ions in the backwashing draw solution influenced the cells on the membrane surface resulting in a reduction in backwashing efficiency. Backwashing with 3 M CaCl2 only removed 40% of the cell surface coverage compared to 93% for backwashing with 3 M NaCl. This study shows that cleaning without the use of harmful chemicals in membrane processes depends strongly on the type of process, the type of fouling, the fouling feed solution chemistry and the initial membrane flux. It also shows that the type of backwashing draw solution is important as even 1 minute of backwashing is sufficient time for interactions to occur between the draw solution and the fouling layer. Finally, this study shows that flux restoration alone cannot indicate how effective membrane cleaning is and that the membrane surface must be examined either by imaging or another surface characterisation method (e.g. SEM-EDS) in order to get a complete picture of cleaning efficiency.