The rejection of endocrine disrupting and pharmaceutically active compounds by NF and RO membranes as a function of compound and water matrix properties

Abstract

This study examined the rejection of 22 endocrine disrupting compounds (EDCs) and pharmaceutically active compounds (PhAC) from raw and 5-μm filtered Lake Ontario water, membrane bioreactor effluent and laboratory-grade water (Milli-Q^®) by ‘loose’ and ‘tight’ nanofiltration (NF) membranes and a reverse osmosis (RO) membrane. Rejection by the ‘tight’ NF membrane was most strongly correlated with compound log K_ow and water solubility with the more hydrophobic compounds showing higher rejection. The RO membrane provided efficient EDC and PhAC removal (>90%). Conversely, in general, the ‘loose’ NF membrane offered poor and variable compound removal. Fouling and the presence of organic matter likely contributed to higher rejections of EDCs and PhACs while higher concentrations of divalent ions appeared to result in lower EDC and PhAC rejection by the membranes.

Introduction

The presence of trace levels (ng/L) of endocrine disrupting compounds (EDCs) and pharmaceutically active compounds (PhACs) in wastewater effluents [1], [2], [3], receiving waters [4], drinking water sources [5] and some treated drinking waters [6], [7] has led to concern about their potential negative impact on ecological and human health [8], [9], [10], [11]. In general, the removal of these compounds by conventional wastewater and drinking water processes has not been shown to be effective [12], [5] and as such there is a need to investigate advanced treatment options, such as activated carbon, advanced oxidation and membrane filtration.

The trace levels (ng/L) of EDCs and PhACs found in the environment combined with analytical limitations, make full-scale EDC/PhAC membrane removal studies challenging. It is often not possible to quantify the rejection of individual compounds either because they are not naturally present in the investigated water at a quantifiable level or their concentrations in membrane permeate are below detection levels. Snyder et al. [13] investigated the removal of 36 EDCs and PhACs during drinking and wastewater treatment processes at pilot- and full-scale and found that reverse osmosis (RO) and nanofiltration (NF) membranes were capable of rejecting most of the studied compounds to below detection levels (<25 ng/L). Drewes et al. [14] examined the removal of 21 EDCs and PhACs in two full-scale RO facilities and observed that, in general, compounds were efficiently rejected to below detection (<40 ng/L) with only caffeine being observed in the permeates (<105 ng/L).

Controlled experiments using bench-scale membrane testing can allow the quantification of rejection efficiency and the investigation of the influence of membrane and compound properties on rejection. A bench-scale study by Košutić et al. [15] examined the rejection of antibiotics from a model wastewater by RO and NF membranes and reported RO and tight NF membranes to offer high rejection (>90%) but the loose NF membrane to offer poor (<40%) rejection of the smaller antibiotic compounds (<200 Da). Van der Bruggen et al. [16] attempted to develop a semi-quantitative method for estimating rejection of organic micropollutants by NF. This model provides an approximation of rejection by taking into account compound molecular weight, hydrophobicity and charge combined with the membrane's molecular weight cut-off (MWCO) and surface charge. Further development of this type of model is needed as molecular parameters including, among others, dipole moment and effective hydrated radius, along with membrane parameters such as pore size distribution, hydrophobicity and charge are not included. It is also important that operational parameters, such as recovery and crossflow velocity, be considered.

Few studies have focussed on the influence of natural water matrices on EDC and PhAC removal by membranes. Yoon et al. [17] studied the removal of 52 EDCs and PhACs by NF and UF membranes and found a decrease in compound rejection with an increase in natural organic matter (NOM) concentration due to competition for membrane adsorption sites. However, this observation may be a result of the filtration tests being run under non-equilibrium conditions and membrane adsorption sites having not yet been exhausted. Schäfer et al. [18] observed a higher rejection of estrone by UF from raw sewage and secondary effluent. Nghiem et al. [19] found that the rejection of estrone and estradiol by NF and RO was enhanced in the presence of organic matter in synthetic waters. Nghiem et al. [20] found that the rejection of bisphenol A, nonylphenol and tert-butylphenol from a synthetic water increased in the presence of 10 mg/L NOM and 10 mM of NaCl. More recently, Nghiem and Hawkes [21] reported that organic fouling can both improve or lessen the retention of PhACs (carbamazepine, ibuprofren, sulfamethoxazole) by NF membranes. They hypothesized that the main mechanisms by which fouling influences membrane rejection are the modification of the membrane surface charge, pore restriction and cake enhanced concentration polarization. Membrane fouling has also been observed to both increase compound adsorption and decrease mass transport causing higher diffusion of solutes across the membrane [22], [23].

In addition, an increase in compound rejection may result from the binding of EDCs and PhACs to NOM due to hydrogen bonding, forming NOM-compound complexes that are larger, have an increased negative charge, and/or a higher affinity for adsorption to the membrane when compared to the compound alone [24], [25], [26]. The presence of cations can also influence the membrane charge and the interaction of compounds and humic acids with each other and the membrane surface [27], [28]. For example, Devitt et al. [26] investigated the rejection of atrazine by NF and UF membranes and observed that atrazine-NOM association decreased in the presence of cations (principally calcium). Plakas et al. [24] studied the removal of atrazine, isoproturon and prometryn by NF and found that the presence of calcium ions alone has a positive effect on pesticide retention but can interfere with the pesticide-NOM complex thus reducing overall retention.

This study examined the rejection of 22 EDCs and PhACs from natural waters by NF and RO membranes. The objective was to examine and compare the rejection of multiple EDCs and PhACs as a function of membrane type and compound properties. The influence of various water matrices on the rejection of EDCs and PhACs by NF and RO membranes was also examined.