Removal of Indicator Micropollutants Included in Directive (EU) 2024/3019 Using Nanofiltration and Reverse Osmosis

Abstract

Contaminants of emerging concern (CECs) comprise a diverse group of substances whose presence in the environment is of increasing concern due to their potential negative effects on human health and the environment. Multiple studies have concluded that nanofiltration (NF) and reverse osmosis (RO) membrane separation mechanisms are effective barriers for organic pollutants, showing generally high removal efficiency. In this study, nine indicator CECs included in the Directive (EU) 2024/3019 concerning urban wastewater treatment were selected and used as a reference to calculate the removal percentage of all micropollutants present in the influent of wastewater treatment plants (WWTPs). According to the regulations, a minimum average removal percentage of 80% of the influent load must be achieved by analyzing at least six out of a set of twelve micropollutants, including those considered in this study. The treatments were conducted using three commercial RO membranes and one commercial NF membrane. Our findings indicate that membrane technology alone can remove over 80% of the micropollutants studied, except benzotriazole. An analysis of the separation mechanisms was carried out to understand the performance of each CEC in relation to each membrane type, taking into account pollutant physicochemical properties and observed removal efficiencies.

1. Introduction

The increasing demand for water and the impacts of climate change are leading to an accelerated reduction in water availability worldwide. The United Nations World Water Development Report 2024 [1] states that almost half of the world’s population faces water scarcity for at least part of the year. Additionally, a quarter of the world’s population experiences extreme water stress, consuming more than 80% of its annual renewable freshwater supply. In response to these challenges, efficient management of existing water resources is crucial, alongside the expansion of desalination and water reuse strategies. It is worth noting that agriculture remains the largest consumer of water, accounting for over 70% of the total water withdrawals, including non-renewable groundwater sources [2].

Moreover, at a global scale, increasing pressure on water resources has led to significant deterioration in water quality, largely due to insufficient or inadequate wastewater treatment. Currently, only 63% of the world’s wastewater is collected and 52% is treated properly [3]. Even when urban wastewater undergoes adequate treatment, major challenges persist due to the presence of CECs in treated water [4,5,6]. Many CECs originate from pharmaceuticals, personal care products, household fungicides, etc., that enter urban wastewater streams.

Research has shown that conventional WWTPs fail to completely remove organic contaminants, including pharmaceuticals, which subsequently persist in surface waters [6,7,8]. According to the NORMAN network [9], a CEC or “emerging pollutant” is “a substance currently not included in routine environmental monitoring programs and which may be candidate for future legislation due to its adverse effects and/or persistence”. The presence of CECs in different environmental matrices is influenced by properties such as volatility, polarity, adsorption capacity, biodegradability, persistence, and hydrophobicity, as well as the characteristics of receiving bodies, including the capacity of water to dissolve them or soil adsorption capacity [10,11,12,13].

Only recently has the removal of CECs in urban WWTPs been prioritized, with regulatory efforts emerging in countries such as Switzerland [5] and within the European Union (EU) [14]. In parallel, the presence of these micropollutants in drinking water [15] or reused water [16] is now subject to increased regulation. The new EU Directive for urban wastewater treatment [17] mandates quaternary treatment processes, requiring at least 80% removal of micropollutants in treatment plants serving urban agglomerations of at least 150,000 p.e. or ≥10,000 p.e. in areas considered sensitive to micropollutant pollution. It is therefore imperative to evaluate cost-effective and technologically feasible micropollutants removal strategies in wastewater treatment. This new directive establishes legally binding limits for a series of emerging contaminants, including pharmaceuticals, personal care products, and other compounds of concern, making their removal a current regulatory and environmental priority. The selection criteria must consider technological, economic and environmental aspects [18]. Among established treatment options, ozonation, activated carbon, and membrane-based technologies are the most widely studied [19].

Membrane processes, although generally more expensive, offer additional benefits, including the simultaneous removal of dissolved salts and microorganisms, which is particularly advantageous for applications such as irrigation of salt-sensitive crops or as the production of drinking water, where high-quality standards are required. In fact, membrane technologies are already widely used in wastewater treatment for reuse as drinking water [20,21,22,23]. Furthermore, membrane technologies do not generate toxic byproducts, although proper disposal of brine concentrates remains necessary. Among membrane separation processes, NF and RO are both technologies capable of removing a wide range of contaminants [4,19,24,25,26]. It is essential to highlight that, although the application of membrane separation technologies for the removal of microcontaminants in wastewater has been extensively studied, the behavior of each CEC with respect to different types of membranes is not uniform and depends on multiple factors. These include the chemical structure of the compound, its physicochemical properties, the dominant separation mechanisms in each membrane (such as rejection by charge, size, polarity, adsorption, etc.), as well as the specific characteristics of the treated wastewater (such as organic load, presence of suspended solids, or competition between compounds). The mechanisms for contaminant rejection are complex [27,28,29], highlighting the importance of understanding how different contaminants interact with the various membrane types available on the market.

The twelve CECs referenced in Directive (EU) 2024/3019 [17] are divided into two groups: category 1, comprising amisulpride, carbamazepine, citalopram, clarithromycin, diclofenac, hydrochlorothiazide, metoprolol and venlafaxine, and category 2, including benzotriazole, candesartan, irbesartan, and a mixture of 4-methylbenzotriazole and 5-methylbenzotriazole (see Table S1 in the Supplementary Materials). For these substances, a minimum average removal efficiency of 80% relative to the influent load of the sewage treatment plant must be achieved. According to the Directive [17], “the percentage of removal shall be calculated on dry weather flow for at least six substances. The number of substances in category 1 shall be twice the number of substances in category 2. If fewer than six substances can be measured in sufficient concentration, the competent authority shall designate other substances to calculate the minimum percentage of removal when it is necessary. The average of the specific percentages of removal of all single substances used in the calculation shall be used in order to assess whether the required 80% minimum percentage of removal has been reached”.

In this study, nine of the twelve micropollutants were selected to evaluate their overall removal efficiency in raw water influent. The micropollutants amisulpride, carbamazepine, diclofenac, metoprolol, and venlafaxine were selected from category 1, and benzotriazole, candesartan, irbesartan, and methylbenzotriazole from category 2. To assess compliance with the Directive’s requirement of 80% minimum removal, the average of the specific removal percentages of all individual substances in the calculation must be considered. The classification, applications, and toxicological risks associated with the studied CECs are summarized in Tables S1 and S2 in the Supplementary Materials.

Research on emerging pollutants and their removal methods in wastewater has gained increasing importance in recent years. However, some of the micropollutants selected for this study have been more extensively studied than others. A bibliometric analysis was conducted using the Scopus online database, covering the period from 1 January 2010 to 31 December 2024. The search was limited as follows: document type “articles”, searched within “title, abstract and keyword” and search documents containing “removal AND wastewater AND (nanofiltration OR osmosis) AND (the name of each CEC of this research)”. The results obtained are illustrated in Figure 1.

Except for carbamazepine and diclofenac, research on the removal of the remaining selected CECs from wastewater using NF or RO membranes remains limited. In the literature reviewed, no study has been found that simultaneously examines the removal of all micropollutants included in the Directive [17].

In this context, the recent adoption of Directive (EU) 2024/3019 [17], which introduces a specific list of priority CECs with binding regulatory targets, requires a rigorous and specific evaluation of their behavior in relation to advanced treatment technologies. Therefore, the present study makes a significant contribution to current knowledge by systematically analyzing the removal of these compounds under controlled conditions, serving as a foundation for their subsequent validation in real treatment scenarios.

The objective of this research is to evaluate the efficiency of NF and RO membranes in removing the indicator micropollutants listed in Directive 2024/3019 [17]. This study aims to ensure compliance with the new regulation, which mandates a global average removal efficiency of 80%.

2. Materials and Methods

2.1. Selection and Characteristics of Membranes

Four different commercial membranes were used: one NF membrane (NF-90 model), supplied by Dupont S.A., Barcelona, Spain [30], and three RO membranes (73AC, 73UAC, and 73HA models), provided by Toray Membrane Spain, S.L. (TMSP), Madrid, Spain [31]. Some applications and advantages of these membrane types are discussed in Table S3 in the Supplementary Materials.

Table 1. Characteristics of the NF and RO membranes used [30,31].

Type of Membrane	Nanofiltration	Reverse Osmosis
Model	NF90-400/34i	73AC	73UAC	73HA
Manufacturer	FilmTecTM (DuPont)	Toray Industries
Type	High removal of nitrates, iron, and organic compounds; low energy consumption	High rejection, chlorine resistant	High rejection, low energy	Extra-low energy
Molecular weight cut-off (MWCO) (Da)	200–400	200	N/A	N/A
Salt rejection (%)	98.7	99.8	99.5	99.3
Maximum operating temperature (°C)	45	45	45	45
Maximum operating pressure (bar)	41.4	41	41	25
Maximum pressure drop (bar)	1.0	1.0	1.0	1.0
pH range in continuous operation	2–11	2–11	2–11	2–11

presents the main characteristics and technical specifications as reported by the manufacturers.

The membranes are designed for effective reuse, have features to combat fouling, are durable, and are expected to have a long service life with proper maintenance. The efficacy of chemical cleanings critically depends on their timeliness: early interventions enable 100% performance recovery, while delays significantly reduce effectiveness. Table S4 in the Supplementary Materials shows the details regarding reuse, fouling, service life, and durability.

2.2. Chemicals and Consumables

Analytical standards of ≥98% purity, sourced from Sigma-Aldrich, Darmstadt, Germany and CymitQuímica, Barcelona, Spain, were prepared in individual dilutions at a concentration of 1 g·L⁻¹ in methanol. The solvents used for standards, working solutions, and analysis included acetonitrile and methanol (both of HPLC quality, supplied by PanReac AppliChem, Darmstadt, Germany), and ultrapure Type I water, obtained from a Milli-Q water purification system (Millipore, Darmstadt, Germany), with its properties being resistivity ≥ 18 mΩ·cm⁻¹ at 25 °C and pH = 6.7.

2.3. Preparation of the Synthetic Matrix

The synthetic feed water was formulated according to DIN 38 412-L24 [32] to a chemical oxygen demand (COD) value of approximately 50 mg O₂·L⁻¹, simulating influent from a low-organic-load urban wastewater treatment plant (WWTP). The detailed composition of the synthetic water is given in

Table 2. Composition of synthetic wastewater (COD_theoretical = 50 mg·L⁻¹).

Compound	Concentration (g·L−1)
peptone	2.6 × 10−2
meat extract	1.8 × 10−2
urea	4.9 × 10−3
MgSO4 7H2O	3.3 × 10−4
KH2PO4	4.6 × 10−3
CaCl2 2H2O	6.6 × 10−4
NaCl	1.2 × 10−3

.

2.4. Experimental Design and Set-Up

Our experimental design, although conducted at lab scale using spiked (doped) water, allows for controlled and reproducible assessment of membrane performance for each individual compound included in the directive.

Each membrane type was tested in duplicate, operating at laboratory scale in continuous mode. The synthetic water was spiked with emerging micropollutants at a concentration of approximately 10 µg·L⁻¹. The system operated at a working pressure of 3 to 4 bar. The use of doped water is a common and scientifically accepted approach to simulate realistic contamination scenarios while maintaining control over concentrations and variability. Moreover, the operational parameters used in our tests were selected to be representative of real-world applications, allowing our results to be easily extrapolated to full-scale systems.

Recently, numerous novel studies have been published at laboratory scale investigating CEC removal in synthetic doped water [33,34] and/or real treated water, enriched or not [34,35,36]. An initial concentration of 10 µg/L was chosen for all target compounds in the spiked water to ensure comparability between substances and remain within environmentally relevant ranges. Numerous studies report significant concentrations of CECs in tertiary treatment effluents from WWTPs. Key detected compounds include amisulpride 0.5 µg/L [37], benzotriazole 1.3 µg/L [37], carbamazepine in ranges from <0.01–2.7 µg/L [35,36,37,38,39], diclofenac 0.4–7.2 µg/L [36,37,39], irbesartan 0.5–0.9 µg/L [35,39], metoprolol 0.03–5.1 µg/L [35,37,39], and venlafaxine 0.1–1.3 µg/L [35,36,37,39]. Several studies have investigated the presence of CECs in surface waters, as these are the primary recipients of treated wastewater effluents that still contain residual concentrations of such compounds [40]. Our research group has also conducted studies on the removal of CEC using spiked water samples with a concentration of 10 µg·L⁻¹ [41,42].

These findings highlight the persistence of CECs even after advanced treatment processes, emphasizing the need to optimize quaternary technologies for effective removal.

For the experiments, a Teflon cell was designed and manufactured to allow continuous filtration through flat sheet membranes. The device enables variation in both the feed pressure and the flow rate of the water to be treated. The water flows transversely across the membrane surface. The system’s maximum operating pressure is 5 bar, and it was operated at 75% conversion.

To control and maintain the working pressure, the system was connected to an Osaka pressure-reading unit. The volume of permeate collected during all experiments was recorded using a precision electronic balance (CB-Complet series, COBOS), which was connected to a computer where mass (g), mass flow rate (g·s⁻¹), and time (s) data were stored for each experiment. The experimental setup is shown in Figure 2.

The membranes were pre-cut to specific dimensions of 18.5 cm in length and 9 cm in width to fit the module, resulting in an active filtration area of 166.5 cm². The hydraulic retention times (HRTs) for each test were approximately 2 h for the NF-90 membrane, 8 h for RO-AC, 7 h for RO-UAC, and 4 h for RO-HA. The permeate flux (LMH) was approximately 1.48 mL·cm⁻²·h⁻¹. Before use, they were conditioned according to the manufacturer’s guidelines: the NF membrane was immersed in ultrapure water for 24 h, while the RO membranes were flushed with ultrapure water for 2 h at a pressure of 4 bar.

2.5. CEC Analysis

Aliquots of feed, permeate, and concentrate were collected in amber borosilicate vials after filtration through hydrophilic PTFE filters with a 0.45 µm pore size and stored at −20 °C until analysis. All samples were analyzed in duplicate to ensure result reproducibility.

The concentration of the CECs was determined using high-performance liquid chromatography coupled to a triple quadrupole mass spectrometer (UHPLC-MS/MS). The systems used was an Agilent 1290 Infinity, Agilent Technologies, Inc., Santa Clara, CA, USA, equipped with an on-line UHPLC interface, and a triple quadrupole spectrometer, Agilent Technologies, Inc., CA, USA, with JetStream and iFunnel technology (UHPLC-1290/QQQ-6490).

A capillary column with a C18 stationary phase (a monolayer of dimethyl-n-octadecylsilane), model Zorbax Eclipse Plus C18, Agilent Technologies, Inc., CA, USA, (2.1 mm × 100 mm × 1.8 µm), was used. According to the manufacturer, this type of column is particularly effective for the separation of acidic, basic, and other highly polar compounds via reversed-phase liquid chromatography. It is suitable for acidic and neutral samples but is especially effective in separating basic compounds that tend to produce poor peak shapes on other types of columns. In addition, these columns are applicable to a wide range of uses and operate within a pH range of 2 to 9, making them compatible with the most commonly used mobile phases.

The mobile phases consisted of (A) Milli-Q water with 0.1 mM formic acid and (B) acetonitrile with 0.1 mM formic acid. The gradient program started with 15% B, increased linearly to 85% B for 10 min, increased at 10.5 min to 98% B (kept 0.5 min), and finally back to initial conditions (15% B) to re-equilibrate the column. The mobile phase flow rate was set at 0.4 mL·min⁻¹, the temperature at 25 °C, and the injection volume at 10 µL. Retention times and MS/MS parameters for each compound analyzed can be found in Table S5 (Supplementary Materials).

2.6. Method Validation

The standards used for calibration curves were prepared from a mixed solution of the micropollutants to be analyzed at a concentration of 1 mg·L⁻¹. A wide range of concentrations was selected, from 0.01 to 50 µg·L⁻¹, in a matrix composed of 80% methanol and 20% water. In all cases, the standards were stored in single-use containers to prevent potential contamination.

For each standard or sample injected into the equipment, chromatograms were integrated semi-automatically using Agilent’s MassHunter Workstation software (Version 10.1), and the peak areas corresponding to the quantification masses of each micropollutant were obtained. Each standard was injected in triplicate. Linearity was assessed over the studied concentration range through regression analysis using the least squares method. Limits of quantification (LOQs) were determined for each compound and were defined as the lowest calibration level meeting two criteria simultaneously: (i) a signal-to-noise ratio greater than 3 and (ii) an accuracy between 80% and 120% based on replicate injections. The values corresponding to the LOQs are presented in Table S6 of the Supplementary Materials.

2.7. Calculation of Removal Percentage

Equation (1) was used to calculate the removal percentage for each CEC. In cases where the concentration measured in the effluent was lower than the LOQ, the LOQ value itself was used to calculate the removal percentage. This approximation ensures that the obtained percentage represents the lowest possible removal value.

$$\mathrm{R}\mathrm{e}\mathrm{m}\mathrm{o}\mathrm{v}\mathrm{a}\mathrm{l}\mathrm{p}\mathrm{e}\mathrm{r}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{a}\mathrm{g}\mathrm{e}={\displaystyle \frac{\mathrm{I}\mathrm{n}\mathrm{f}\mathrm{l}\mathrm{u}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}-\mathrm{E}\mathrm{f}\mathrm{f}\mathrm{l}\mathrm{u}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}{\mathrm{I}\mathrm{n}\mathrm{f}\mathrm{l}\mathrm{u}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}}·100,$$

(1)

For the calculation of global removal, the average of the specific percentages of removal of all single substances was used.

3. Results and Discussion

3.1. Elimination of CECs

Figure 3 shows the removal percentages of each micropollutant for the membranes studied. The horizontal green line shows 80% removal.

As observed, all contaminants, except benzotriazole, exhibit removal rates exceeding 80% with at least one type of membrane. For methylbenzotriazole, the removal efficiency was relatively modest. As expected, RO membranes retain a higher number of contaminants compared to NF. Among the three RO membranes evaluated, the AC membrane achieved the highest removal percentage. This is because the UAC and HA membranes, which are classified as low-energy and extra-low-energy membranes, respectively, operate at lower pressures, thereby reducing energy consumption. However, this design also slightly decreases removal efficiency. Nevertheless, the NF membrane demonstrates significant contamination removal capability.

3.2. Analysis of CEC Rejection

Mechanisms involved in the removal of micropollutants include size exclusion, membrane affinity, electrostatic interaction, and interaction with macromolecules in solution [43]. In the case of RO, very recent studies [44,45,46,47] have shown that the transport of water and solutes across the membrane is not governed by the solution/diffusion model [48,49] but is more likely explained by a solution/friction mechanism, in which both water and dissolved salts flow through pores within the membrane structure. The active layer of the membrane consists of a polymeric network with continuous subnanometric channels/pathways filled with water, where solutes are dispersed in the aqueous phase. Ions partition into the membrane through three mechanisms: steric exclusion, the Donnan effect, and dielectric exclusion. Ion flux through the membrane is driven by the chemical potential gradient, which includes contributions from concentration and electric potential gradients, as well as ion advection via the permeate flux [44].

For micropollutants, partitioning between the influent and the membrane is determined by steric exclusion, hydrophobic interaction, hydrostatic repulsion, and interaction with the fouling layer [50,51]. Consequently, the analysis of pollutant separation using the studied membranes must be correlated with key physicochemical properties of the pollutants, such as molecular weight, the octanol/water partition coefficient (which determines the hydrophilic nature of each compound), ionic charge, or dissociation constant.

The most relevant properties of the contaminants studied are listed in Table S7 of the Supplementary Materials. This information was obtained from databases such as PubChem and DrugBank. The molecular weights of the analyzed CECs range from 119.1 g·mol⁻¹ (benzotriazole) to 440.5 g·mol⁻¹ (candesartan). This property is crucial in membrane technologies, as it influences steric exclusion. Water solubility varies from insoluble (amisulpride) to highly soluble (metoprolol). This property, along with the octanol/water partition coefficient (log K_ow), which ranges from 1.06 (amisulpride) to 6.1 (candesartan), determines the compound’s affinity for water, with amisulpride being the most hydrophilic compound and candesartan the most hydrophobic. Membrane surfaces are often negatively charged, which favors the repulsion of anions and the attraction of cations, thereby influencing selectivity. At neutral pH, the studied micropollutants exhibit a variety of ionic charges: positive charge (amisulpride, metoprolol, and venlafaxine), neutral charge (benzotriazole, carbamazepine, and methylbenzotriazole), and negative charge (candesartan, diclofenac, and irbesartan). The acid dissociation constant (pk_a = −log k_a) ranges from 2.45 (candesartan) to 13.9 (carbamazepine), measuring the compounds’ ability to dissociate in aqueous solutions. A high k_a value corresponds to a strong acid, as it readily dissociates into its ions, while a low k_a value is associated with a weak acid, which dissociates to a lesser extent.

Among all physicochemical properties influencing the separation of micropollutants, size exclusion and adsorption have been identified as the dominant removal mechanisms [51]. Figure 4a illustrates the relationship between removal efficiency and the molecular weight of the analyzed CECs, while Figure 4b shows the correlation between removal efficiency and the log K_ow of each CEC. In both cases, the graphs were designed to highlight the decreasing trend in removal efficiency as a function of the evaluated physicochemical properties.

Benzotriazole is the micropollutant with the lowest removal percentage, which can be attributed to its low molecular weight of 119 g·mol⁻¹ (Figure 4a), as well as its low log K_ow value of 1.44 (Figure 4b), indicating a high affinity for water. Both factors significantly facilitate its passage through the membrane, hindering efficient removal, especially in the NF membrane, which has a larger mean pore diameter than the RO membranes.

In contrast, the CECs with the highest removal percentage across all membranes are candesartan and irbesartan, with the highest molecular weights (440 and 429 g·mol⁻¹, respectively). More than 80% of diclofenac is removed with all membranes studied, despite its intermediate molecular weight of 296 g·mol⁻¹, due to its high hydrophobic character (log K_ow = 4.51), which enhances its retention.

For carbamazepine, the dominant exclusion mechanism appears to be its size, as its molecular weight of 236 g·mol⁻¹ allows RO membranes to retain more than 80%, while NF membrane retention is lower. Amisulpride has the highest affinity for water (log K_ow = 1.06), which may hinder its retention. However, its positive charge enhances electrostatic attraction to the negatively charged membrane surface, while its relatively high molecular weight (369.5 g·mol⁻¹) favors molecular exclusion. The interplay of these properties results in relatively high removal percentages but with variations depending on the membrane type.

Metoprolol and venlafaxine are effectively eliminated only by the RO-AC membrane. These two compounds share similar characteristics, including positive ionization, which complicates retention, intermediate molecular weights (260 and 280 g·mol⁻¹, respectively), and a relatively high affinity for water.

3.3. Global Removal of CECs

According to the Directive [17], to determine whether the minimum removal percentage of 80% has been achieved, the calculation must be based on the average specific removal percentages of all individual substances. Figure 5 shows the average specific percentage of removal for the nine selected CECs for each membrane.

The global average removal efficiencies obtained for each membrane are 68% for NF-90, 91% for RO-AC, 78% for RO-UAC, and 78% for RO-HA.

Thus, in the unlikely event that none of the micropollutants undergo removal during the pretreatment processes in the WWTPs, achieving the 80% removal rate set by the Directive [17] would be feasible by treating the water with the RO-AC membrane.

In general, the physicochemical and biological processes occurring in WWTPs contribute to the removal of micropollutants, with removal efficiencies that vary widely [52,53,54,55]. Several studies have analyzed the removal efficiencies of specific micropollutants addressed in this investigation, such as venlafaxine, carbamazepine, metoprolol, diclofenac, and candesartan, in various treatment plants [6,52,56]. Among these, carbamazepine has been identified as particularly resistant to biodegradation, with removal efficiencies in WWTPs ranging from 30 to 40% [53,57]. Diclofenac, one of the most prevalent pharmaceutical compounds in the water cycle [58], has shown removal efficiencies in WWTPs ranging from 9% to 76% [8,52,53,55]. Other studies have reported removal efficiencies of 31–39% and 10–78% for venlafaxine and metoprolol, respectively [52,54,59]. Thus, some degree of removal of the micropollutants listed in the Directive [17] is expected to occur during conventional wastewater treatment. Consequently, this removal must be considered when calculating the total required removal rate of 80%. In practical terms, the global average removal efficiency achieved through quaternary treatment should supplement the removal occurring in WWTPs to ensure that 80% of the pollutant load entering with the influent wastewater is eliminated.

Although NF membranes have proven effective in the removal of some micropollutants and offer operational advantages such as lower energy consumption [60,61,62,63], their efficiency is highly dependent on the specific characteristics of each compound. In this study, it was observed that two azole compounds exhibited the lowest removal efficiencies through NF. This behavior aligns with what has been reported in the literature, where their high resistance to conventional and even advanced treatment processes, such as advanced oxidation, has been documented [64]. This persistence is primarily attributed to the presence of triazole and/or imidazole functional groups in their molecular structure, which hinder their degradation. Therefore, while NF is effective for some compounds, it is not sufficient on its own for the effective removal of all the emerging contaminants considered as indicators in the Directive [17], highlighting the need to explore combined treatments or more specific strategies.

3.4. Challenges in the Treatment of Real Effluents and the Influence of TDS on Membrane Treatment

Although our experiments were conducted using spiked water to ensure precise control over the concentration of CECs, we recognize that the treatment of real effluents presents several challenges. In real-world conditions, the effluent is composed of a variety of contaminants, including not only the CECs but also organic matter, suspended solids, and other compounds that can interfere with the efficiency of membranes. The presence of suspended solids and organic matter in real effluent can increase fouling rates, which reduces membrane efficiency and increases operational costs due to the need for frequent cleaning and maintenance. Therefore, while the results obtained in this study under controlled conditions provide a solid foundation for evaluating membrane technologies for CEC removal, further validation under real treatment conditions is required, taking into account the specific factors of the effluent that may influence the efficiency of membrane technology.

Furthermore, the interaction between CECs and other contaminants in real effluent can influence the separation mechanisms, making the removal of CEC less efficient than under controlled laboratory conditions. For instance, competition between different contaminants for adsorption sites on the membrane can reduce the effectiveness of treatment.

Regarding the influence of total dissolved solids (TDS) on membrane performance, it is known that elevated levels of TDS can alter both membrane permeability and the interaction between CECs and the membrane. This has been discussed in the literature, as higher TDS concentrations can lead to changes in the physicochemical properties of both the membrane and the contaminants, influencing the efficiency of the separation process [65].

A key advantage of our study is its focus on treated wastewater, which generally tends to have low conductivities (salt concentrations), especially in the case of biologically treated water. This factor facilitates membrane treatment, as lower TDS concentrations reduce the risks associated with membrane fouling and can improve separation efficiency. Although the presence of high TDS concentrations can alter membrane permeability and the interaction between CECs and membranes (as discussed previously), treated wastewater presents a less aggressive environment for membranes in terms of the accumulation of salts and minerals. This is especially relevant in contexts where the treated water is not influenced by brackish well water or seawater sources, which typically have higher TDS concentrations.

4. Conclusions

Most of the studied CECs have individual removal rates of more than 80% with at least one selected membrane type, except for benzotriazole, due to its low molecular weight and affinity for water.

With the RO-AC membrane, removal rates exceed 80% for all micropollutants, except for benzotriazole, which only reaches 67%.

The RO-AC membrane showed the highest overall removal efficiency among all the compounds analyzed, including the most persistent ones. Although the RO-UAC and RO-HA membranes also performed well for certain compounds, their performance was less consistent.

The results demonstrate that, while NF membranes offer advantages in the removal of some micropollutants, they have limitations when it comes to certain azole compounds. In this regard, although NF is effective for some compounds, it is not sufficient on its own for the effective removal of all CECs considered as indicators in the Directive [17], suggesting the need to consider combined treatments or complementary approaches.

The presence of TDS in water significantly influences membrane performance, as high concentrations can alter membrane permeability and its interaction with CECs. However, treated wastewater, with low TDS concentrations, presents a more favorable environment for membrane treatment, minimizing fouling risks and improving process efficiency.