Victor Yangali Quintanilla awarded doctoral degree

Now and in the future, the ever-growing demand for drinking water will lead many cities to implement indirect water reuse programs, where wastewater effluent becomes part of the drinking water sources. Pollution of those sources with emerging contaminants (micropollutants) such as endocrine disrupting compounds, pharmaceutically active compounds, pesticides and personal care products is a fact known worldwide.

Although the risks of micropollutants in sources of water are partly recognised, interpretation of consequences are controversial; thus, the future effects of altered water with micropollutants remains uncertain and may constitute a point of concern for human beings when potable water consumption is involved. Therefore, many drinking water utilities target as an important goal high-quality drinking water production to lessen quality considerations that may arise from the consumers.

In this thesis, nanofiltration (NF) and reverse osmosis (RO) are demonstrated to be appropriate technologies for removing a large number of micropollutants; however, the performance of NF and RO can be questioned because there are limited tools that optimise quantification of the removal of contaminants. Therefore, in this research, by means of the use of multivariate data analysis techniques, removal quantification is effectively determined and more understanding of the separation of micropollutants by membranes is achieved.

The removal of micropollutants (1,4-dioxane, estrone, atrazine, bisphenol A, 17β-estradiol, 17α-ethynilestradiol, ibuprofen, sulfamethoxazole, naproxen, fenoprofen, ketoprofen, phenazone, carbamazepine, caffeine, acetaminophen, metronidazole and phenacetine) from synthetic water solutions and surface water was investigated in filtration experiments using NF membranes. Two aromatic polyamide nanofiltration (NF-90 and NF-200) membranes were utilized in laboratory-scale experiments. Additionally, other NF membranes and RO membranes were also studied, taking into account data (of removal of micropollutants) generated during research conducted in other laboratory, pilot- and full-scale installations.

Physicochemical properties of organic compounds were used to compare similar or different removals achieved by clean and fouled membranes. High rejections (90–99%) were observed for ionic compounds compared to neutral compounds (hydrophilic and hydrophobic, 20–99%).

It was demonstrated that electrostatic repulsion between the negative charge of the ionic species of the solutes and the negative charge of the membrane surface was the main mechanism of rejection for ionic compounds. On the other hand, for hydrophilic neutral and hydrophobic neutral compounds, the rejections were mainly related to variables of size in the order of length > effective diameter > equivalent width > molecular weight.

Hydrophilic neutral compounds were less rejected by an NF membrane when their size (length or equivalent width) is less than or close to the pore size of the membrane (< 1nm). By contrast, hydrophobic neutral compounds are influenced by log Kow (hydrophobicity) contributing to partitioning when the pore size of the membrane is close to the size of the compound. Partitioning of a hydrophobic neutral compound (log Kow > 3) will occur after adsorption of the solute into/onto the membrane until saturation.

A quantitative structure-activity relationship (QSAR) model for predicting the rejection of pharmaceuticals, endocrine disrupters, pesticides and other organic compounds by NF membranes was developed. Principal component analysis, partial least-square regression and multiple linear regressions were used to find a general QSAR prediction model that combines interactions between membrane characteristics and compound properties.

Subsequently, the development of an extended model applicable to NF and RO was realized. Prediction of the rejection of neutral organic compounds by polyamide NF and RO membranes was evaluated using artificial neural network (ANN) models. The ANN models were developed based on QSAR equations.

Finally, the results obtained in this study and other existing information were used to demonstrate that NF is an effective barrier against pharmaceuticals and endocrine disrupters. It raises the question of why RO is used in existing water reuse facilities when NF may be a more cost-effective and efficient technology able to tackle the problem of emerging organic contaminants.

Sufficient support information and, more importantly, experimental and current practice results provide proof that NF should be considered instead of RO in future water reuse projects. It was demonstrated that NF can comply with current regulations of water quality regarding pesticides removal, and its compliance with future regulations for pharmaceuticals and endocrine disrupters may be realistic.

Specifically, when there is a potential presence of difficult to remove organic contaminants such as NDMA and 1,4-dioxane, implementation of additional processes in a wastewater treatment plant will help to remove these compounds and reduce their presence during further water treatment with intentions of indirect reuse. Biodegradation of NDMA can be achieved through aquifer recharge and recovery, and other treatment techniques that are less expensive than advanced oxidation processes can help to remove 1,4-dioxane.