Degree Name

Doctor of Philosophy


School of Civil, Mining, and Environmental Engineering


A comprehensive assessment focusing on the rejection of eight N-nitrosamines by reverse osmosis (RO) membranes was conducted for planned potable water recycling applications. The effects of feed solution characteristics, operating conditions, fouling, chemical cleaning and membrane characteristics on the rejection of N-nitrosamines were first examined at laboratory scale. Field sampling campaigns were carried out at full-scale water recycling plants to provide longitudinal and spatial insights to the rejection of N-nitrosamines. For the prediction of the rejection of N-nitrosamines by spiral-wound RO membrane systems, a mathematical model was developed based on the irreversible thermodynamic principle and hydrodynamic calculation. The model was validated with the results obtained from a pilot-scale RO system.

The results reported here indicate that the rejection of N-nitrosamines by a given membrane increased in the order of increasing molecular weight, suggesting that steric hindrance was a dominating rejection mechanism of N-nitrosamines. The results also indicate that pH, ionic strength, and temperature of the feed solution can exert an influence on the rejection of N-nitrosodimethylamine (NDMA) and in some cases other N-nitrosamines. An increase in the feed temperature led to a significant decrease in the rejection of all N-nitrosamines and the impact was more pronounced for the small molecular weight N-nitrosamines. In general, the rejection of N-nitrosamines increased when the membranes were fouled by tertiary effluent. The rejection of low molecular weight N-nitrosamines was most affected by membrane fouling. From the results reported here, it appears that low molecular weight foulants present in the tertiary effluent can restrict the solute pathway within the active skin layer of membranes, resulting in the observed increase of solute rejection. Caustic chemical cleaning resulted in an increase in membrane permeability but caused a notable decrease in the rejection of N-nitrosamines. The impact of caustic cleaning was not permanent and could be significantly reduced by a subsequent acidic cleaning step. In general, the rejection of NDMA and N-nitrosomethylethylamine (NMEA) increased with decreasing membrane permeability. The impact of membrane permeability became less important for higher molecular weight N-nitrosamines. In addition to the mean free-volume hole-radius of RO membranes which was measured by the positron annihilation lifetime spectroscopy (PALS), other membrane parameters and properties such as the free-volume hole-radius distribution and thickness of the active skin layer can also play a role in governing the rejection of small and uncharged solutes by RO membranes.

During the sampling campaigns at three full-scale water recycling plants, NDMA was detected in all RO feed samples varying between 7 and 32 ng/L. Overall rejection of NDMA among the three RO systems varied widely from 4 to 47%. Data presented here suggest that the feed temperature can influence rejection of NDMA. A considerable variation in NDMA rejection across the three RO stages (14-78%) was also observed. Overall N-nitrosomorpholine (NMOR) rejections were consistently high ranging from 81 to 84%. On the other hand, overall rejection of N-nitrosodiethylamine (NDEA) varied from negligible to 53%, which was considerably lower than values reported in previous laboratory-scale studies.

The developed model was able to accurately describe the rejection of N-nitrosamines under a range of permeate flux and system recovery conditions. The modelled Nnitrosamine rejections were in good agreement with values obtained experimentally using a pilot-scale RO filtration system. The modelling results also revealed that an increase in recovery caused a decrease in the rejection of these N-nitrosamines, which is consistent with the experimental results. Further modelling investigations suggested that NDMA rejection by a spiral-wound system can drop from 49 to 35% when the overall recovery increased from 10 to 50%. The model developed from this study can be a useful tool for water utilities and regulators for system design and evaluating the removal of N-nitrosamines by RO membranes.