Potential of Coagulation, Chlorine Dioxide Oxidation and Sand Biofiltration in Quaternary Treatment of Municipal Wastewater

Abstract

Removal of micropollutants from wastewater is currently drawing a lot of attention in the field of municipal wastewater treatment plants. Firstly, this is because of their unpredictable and potentially damaging fate in the environment, and secondly, due to newly established requirements in the relevant European Union Directive (EU) 2024/3019. This article assesses coagulation, oxidation with chlorine dioxide, sand biofilter, and their combinations as potentially cheaper and sustainable alternatives to well-established, but more expensive methods. For the experiments, citalopram, carbamazepine, and diclofenac were chosen as representatives of micropollutants. Removal efficiencies were evaluated using HPLC, COD, and absorbance UV/VIS at different wavelengths. The demand for chlorine dioxide was assessed using the chlorophenol red method. Owing to analytical limitations, the concentrations examined were in mg/L, which significantly exceeds actual concentrations found in wastewater. The application of stand-alone chlorine dioxide oxidation exhibited the best performance as it sufficiently removed citalopram and diclofenac. On the contrary, biodegradation was found to be the least efficient method, as none of the compounds tested were sufficiently removed in a short period of time. However, the results may be partially biased owing to high concentrations of the micropollutants assessed. In the following stage of the research, the evaluation of transformation products is desired to prevent such potentially harmful chemicals from entering the environment.

1. Introduction

Micropollutants, such as pharmaceuticals, personal care products, pesticides, and different additives, are frequently used in our daily lives, and their resulting occurrence in the environment leads to increasing concerns [1]. Micropollutants are present in the environment and wastewater in trace concentrations (µg/L or even ng/L) [1,2,3]. However, they can still affect human health and the quality of the environment due to their high toxicity, persistence, and/or bioaccumulation [3]. In many cases, the hazard of particular compounds is not fully clear yet, but their input to the environment is already significant enough to call for our attention [4].

As a result, the reduction of the pollutants released into the environment, as well as their removal from the environment, has become topical in different fields of technological processes, including wastewater treatment. Wastewater treatment plants (WWTPs) are a significant source of micropollutants in the environment, as conventional WWTPs can only remove 30 to 65% of the micropollutants detected [2]. Additionally, a new European Union Directive (Directive (EU) 2024/3019) requires monitoring and at least 80% removal of certain compounds [2,5]. Even though this removal rate applies to the whole wastewater treatment process, due to the lack of information about the micropollutants’ concentration degradation at the tested WWTP and unrealistic concentrations of micropollutants used in this work, this article assesses the general potential of several methods and evaluates, by the EU Directive requirements, only the quaternary stage.

This work examines three compounds from the EU directive list—namely carbamazepine (CBZ), citalopram (CIT), and diclofenac (DCF), all from Category 1 [5]. The substances were selected based on the availability of the compound standards.

The introduction of the fourth stage of treatment should also enable wider reuse of treated water, which could then be used for crop irrigation without concerns for micropollutants being absorbed from the water into the crops and the food chain [5]. This kind of water reuse may have a strongly positive impact not only on the regions suffering from water scarcity, but it would also generally lead to water sustainability [6]. To emphasize the need for it, Regulation 2020/741 of the European Parliament and the Council sets minimum requirements for water reuse based on the wastewater effluent quality for both the crops consumed raw and technical or energy crops [2].

Possible methods of so-called quaternary treatment are usually readily present; however, they are more commonly used in water treatment processes or for microbial disinfection at WWTPs and not for the wastewater treatment process as such [7]. Therefore, an appropriate assessment is desired concerning their potential use in micropollutant removal and new arrangements within the treatment processes established. As the efficiency of each method is limited, optimal solutions typically combine several techniques [8].

Widely proposed methods include adsorption (e.g., on activated carbon or biochar), separation on a membrane, coagulation, oxidation by AOPs, and biodegradation [3,4,8]. The Swiss wastewater treatment practices, for instance, take into account only ozonation, powdered activated carbon, granular activated carbon, or their combination [9]. As of today, these are the main technologies that seem feasible for large-scale applications [1].

The application of activated carbon is characterized by lower investment and operating costs, compared with other possibilities, and it can be effectively combined with other processes (absorbing products of prior degradation). However, the regeneration and disposal costs of used carbon and generated sludge are not negligible, and the effluent may still include compounds in their unchanged forms. Also, the implementation of preceding steps to reduce effluent organic matter as much as possible may be necessary to prolong the operating lifetime of carbon [10,11].

The advantages of AOPs, typically ozonation, are the destruction of the pollutant (in contrast with the activated carbon, where the micropollutants are concentrated on the media and a hazardous sludge is produced), applicability even in high pollutant concentrations, easy operation, high efficiency, and fast reaction [3,10]. Also, in combination with other processes, the efficiency and economic feasibility of AOPs may increase even further. In contrast, the disadvantages consist of the frequent production of toxic compounds, non-selectivity, difficult scalability, and high investment and operational costs [4,10].

A number of options are available for the realization of quaternary treatment. However, deep-bed biofilters with granular activated carbon or sand loading are the most common [10]. When combined with ozonation, powdered activated carbon serves as a catalyst, enhancing the generation of ozone and limiting mineralization while preserving its role as an adsorbent of the pollutants [4].

Currently, these promising techniques are quite costly and could increase the treatment expenses by several tens of percent [12,13,14]. Also, the requirements for energy and materials tend to be high, which contradicts their sustainability. With respect to that, this article focuses on alternative methods in the context of micropollutant removal, namely oxidation with chlorine dioxide, coagulation, and biofilm degradation.

1.1. Coagulation

By adding a coagulating (and flocculating) agent, flocs containing micropollutants are formed. As these flocs sediment, they enable the reduction of the pollutants in treated water [2,15]. Nevertheless, the efficiency of micropollutant removal hardly exceeds 50% [16]. On the other hand, the presence of coagulable particles can worsen the performance of subsequent methods, and therefore, coagulation and/or flocculation should be considered as an effective means of pretreatment [16].

1.2. Ozonation

The best-known AOP is ozonation. It is also well-established as a micropollutant-removal step in Switzerland and Germany [10]. Despite its popularity, ozone formation has a very low conversion rate (up to only 8%), and the majority of energy input (ca. 85%) is wasted in the form of heat [4]. Applying ozone can also lead to the formation of undesired byproducts of oxidation [17]. Owing to that, after ozonation, a post-treatment consisting of biodegradation of byproducts (especially those potentially harmful) is usually implemented [10].

1.3. Chlorine Dioxide

Chlorine dioxide (ClO₂) is commonly used in water treatment instead of Cl₂. Unlike Cl₂, it exhibits the limited production of halogenated disinfection byproducts, and its oxidative capacity is 2.5× greater than Cl₂ on both a molar and weight basis [3].

ClO₂ is one of the possible oxidants in AOPs, combining broad-spectrum disinfection with effective oxidation of organic pollutants [18,19]. When applied to remove micropollutants, it interacts especially with those including functional groups, such as aniline, phenolic, aromatic, or tertiary amine groups, which are all rich in electrons [3].

The advantages of the utilization of ClO₂ lie in its strong oxidizing effect, the low formation of disinfection byproducts, and fairly easy implementation in treatment processes [3]. For example, Tampere Water in Finland has operated with ClO₂ oxidation preceded by coagulation and followed by sand and granular activated carbon filtration for more than 20 years [19].

With respect to its short half-life and incompressibility, ClO₂ must be generated on-site [20]. Both main ClO₂ generation and application products, chlorite and chlorate, may be dangerous to human health [18,20]; therefore, their concentration is limited, for example, in drinking water [18].

Drawbacks of ClO₂-based processes may be, for instance, the need to deal with chemical residues in treated water and rather significant costs due to the demand for energy or chemicals [18].

Sulfite is usually used to quench redundant ClO₂ to stop oxidation and disinfection, and limit its possible negative effect on the environment. However, the radicals formed can react with other molecules present in the system, potentially enhancing micropollutant removal [20].

Cassol et al. [21] found that the addition of Fe²⁺ coagulant after ClO₂ preoxidation in the particular system could effectively reduce ClO₂⁻ and ClO₃⁻ generated during the processes, increase the concentration of radicals, and therefore lead to better micropollutant removal. However, once Fe²⁺ and organic matter create a complex, it starts to act anti-oxidatively, and an additional alternative method (e.g., coagulation or adsorption on activated carbon) must be introduced [18].

ClO₂ is also one of the agents used to control biofilm growth. Nevertheless, certain cultures may be harder to moderate, and co-culture samples seem to be more resistant to oxidation than monocultures [22].

To the best of our knowledge, this study is the very first to assess the potential of ClO₂ in quaternary treatment.

1.4. Biodegradation

Biodegradation is sustainable, environmentally friendly, and, compared to other methods, it is also a low-cost method of micropollutant removal. In conventional wastewater treatment, it is the dominant mechanism of (micro)pollutant elimination [4]. Large organic molecules are decomposed into less complex organic or even simple inorganic substances, such as CO₂, H₂O, and others that are not harmful to the environment [4].

Biofiltration, as such, combines filtration and biological degradation. Microorganisms in a biofilm attached to filter media use organic matter as a source of energy or nutrients to multiply. The concentration of substrate, oxygen, and other nutrients affects the prosperity of microorganisms. Compounds occurring in higher concentrations tend to be used as an energy source, while those with minor concentrations are processed in co-metabolic pathways. The most common filter materials are sand, anthracite, and granular activated carbon [23].

In their study, Paredes et al. [24] confirm the convenience of the combined method, as the presence of the microbial fraction was found to significantly enhance the micropollutant removal performance in most cases [24].

The quality of the influent is one of the main factors affecting microbial growth. It must contain C, N, and P (for heterotrophic bacteria, preferably in a molar ratio of 100:10:1), but also certain micronutrients. Since C is usually the limiting nutrient, the addition of easily degradable substrate may be helpful [23].

1.5. Economic Evaluation

Ianes et al. [12] reviewed and compared capital (CAPEX) and operational (OPEX) expenses connected with ozonation, PAC, and GAC quaternary processes, as these particular processes seem to be the most promising in terms of their efficiency in micropollutant removal. For ozonation, annual costs per PE ranged from 1.2 to 6 € depending on the applied O₃ concentration (considering 200 L/(day∙PE), it equals 0.016–0.082 €/m³). For GAC, the annual costs ranged from 3 to 12 €/(PE∙y) depending on the regeneration time (0.032–0.164 €/m³). In the case of PAC, the dosage and disposal of the substance were the crucial factors influencing the cost—from 7.5 to 13.5 €/(PE∙y) (0.100–0.180 €/m³). The article mentioned calculated the CAPEX assuming a 20-year process lifetime [12].

To put this in context, in another study, Moral Pajares et al. [14] mention the prices for the treatment of wastewater at a conventional WWTP ranging from 0.17 to 0.53 €/m³. Additionally, it is also important to consider certain cost increases stemming from inflation throughout the years.

As could be seen, the choice of the quaternary treatment would significantly affect the cost of the overall water treatment, and therefore, potentially cheaper alternatives should also be considered.

2. Experimental Part

The experimental part of this work applied the processes of coagulation pretreatment and chlorine dioxide oxidation to a real sample of a municipal WWTP effluent and effluent samples spiked with three micropollutants—CIT, CBZ, and DCF. A small continuous biofilm reactor was operated with media circulation using a biomass mineral carrier (sand). Subsequently, long-term kinetic tests were conducted, focusing on the biodegradation of the micropollutants and the use of adapted biomass. Also, the performance of the biological system was assessed.

All experiments were conducted at laboratory temperature (18–23 °C). The pH in water tested for biodegradation was not adjusted, as it complied with the effluent requirements (pH 6–9). In the coagulation experiments, the pH value decreased to up to 3.7, which was not adjusted in these experiments, as this is a known phenomenon in wastewater treatment practice and is adjusted right before the effluent to comply with the abovementioned requirements. The oxidation and spiked coagulation experiments were performed in triplicate.

2.1. Materials and Chemicals

2.1.1. Wastewater Treatment Plant Effluent

WWTP effluent water was taken from the outflow stream at a municipal WWTP with approximately 70,000 population equivalent (PE). The technological line of the secondary treatment of this plant consists of regeneration, denitrification, and nitrification stages, with the simultaneous precipitation of phosphorus; no tertiary or quaternary stages are present. The effluent fulfills all the requirements given by the Czech legislation. Water from the effluent stream was not obtained periodically but based on the need for a real matrix to run the sand filter model as described in Section 2.5.1 or to conduct experiments with micropollutants. The average characteristics of the effluent were as follows: COD_Cr 29.79 mg/L, BOD 4.33 mg/L, TSS 3.88 mg/L, TN 13.73 mg/L, TP 1.05 mg/L, and pH 7.3. These values were considered as reference values for subsequent conclusions.

2.1.2. Chlorine Dioxide

Synthesis

To generate a ClO₂ solution with a concentration of 3 g/L in demineralized water, the following reaction was used as described in Equation (1) [25].

$$5 {\mathrm{N}\mathrm{a}\mathrm{C}\mathrm{l}\mathrm{O}}_2+4 {\mathrm{N}\mathrm{a}\mathrm{H}\mathrm{S}\mathrm{O}}_4\to 4 {\mathrm{C}\mathrm{l}\mathrm{O}}_2+4 {\mathrm{N}\mathrm{a}}_2{\mathrm{S}\mathrm{O}}_4+\mathrm{N}\mathrm{a}\mathrm{C}\mathrm{l}+{2 \mathrm{H}}_2\mathrm{O}$$

(1)

Stoichiometrically, 5 g of crystalline NaClO₂ and 7 g of crystalline NaHSO₄ (in slight surplus) were first dissolved separately in demineralized water. The NaClO₂ solution was quantitatively transferred to a 1 L volumetric flask half-filled with demineralized water. Then, the NaHSO₄ solution was slowly (in several rounds) added to the flask, and the flask was filled to 1 L with demineralized water. Finally, the closed flask was placed in a fridge overnight.

Analysis

The concentration of ClO₂ in the stock and diluted working solutions was determined by absorbance at 360 nm and calculated according to Slovak standard STN 75 7151 [26].

The analysis of ClO₂ concentration in experiments was performed with chlorophenol red (CPR), again as described in Slovak standard STN 75 7151 [26]. The main difference from the standard was the higher-mentioned preparation of ClO₂. In all cases, the CPR procedure was performed as described in the standard for concentrations of 0–1 mg/L of ClO₂. The concentration of ClO₂ was then calculated from the calibration made in demineralized water and the derived equation, linking the decrease in absorbance at 570 nm and the known concentration of ClO₂.

2.1.3. Selected Micropollutants

The analytes used were as follows:

• Citalopram Hydrobromide—CAS 59729-32-7, Sigma-Aldrich (St. Louis, MO, USA), purity ≥ 98% (HPLC), solid;
• Carbamazepine—CAS 298-46-4, Sigma Aldrich, purity ≥ 98% (HPLC), powder;
• Diclofenac—CAS 15307-86-5, Sigma Aldrich, purity ≥ 98% (HPLC), powder.

The selection was made from the list of micropollutants provided by the EU Directive and the current availability of the chemicals.

Compound Characteristics Influencing Removal Mechanisms

Physicochemical properties of compounds notably affect their removal efficiency.

The sorption is significantly dependent on hydrophobicity, acidity, and the solid–water distribution coefficient (K_d) of a compound. The rule of thumb for estimating the adsorption potential says that the logarithm of the octanol–water partition coefficient (logK_ow) under the value 2.5 indicates low potential, 2.5–4.0 is moderate potential, and values above 4.0 show high potential. If the acid dissociation value (pKa) is lower than the pH, the adsorption may be more difficult due to electrostatic repulsion. The lower the logK_d value falls, the less significant sorption (under 2.48—insignificant sorption) can be noted [16].

As for biodegradation, substances with long, widely branched chains, saturated bonds, polycyclic structures, and electron-withdrawing functional groups (sulfate, halogen, etc.) are rather persistent. On the other hand, linear compounds with short side chains, unsaturated bonds, and electron-donating functional groups are generally easier to degrade [16].

The key physicochemical factors of the compounds tested are listed in

Table 1. Physicochemical Properties of Subjected Compounds [27,28,29].

Compound	Therapeutic Group	Molecular Formula	MW	pKa	logKow	logKoc	BCF
CBZ	anticonvulsant, analgesic drug	C15H12N2O	236.27	13.9	2.45	2.71	ca. 15
CIT	antidepressant	C20H21FN2O	324.4	9.78	1.39	5.63	ca. 40
DCF	nonsteroidal anti-inflammatory drug	C14H11Cl2NO2	296.15	4.2	4.51	2.39	ca. 3

Key: MW—molecular weight, pKa—acid dissociation, K_ow—partition coefficient of octanol–water, K_oc—partition coefficient on organic carbon (directly proportional to solid–water distribution coefficient K_d), BCF—bioconcentration factor.

, together with their therapeutic applications.

All three micropollutants were used in their powder/crystalline form and dissolved in effluent water following the needs of the experiment. The initially intended concentration of 4 mg/L was not reached due to poor solubility in water. Eventually, the most convenient dose resulted in being 2 mg/L.

The concentration applied significantly exceeded the concentrations of micropollutants detected in real wastewater samples. This change was made due to a quite high detection limit of the used analytical method (see Section 2.2.3) and operational limits of the laboratory for spiking the samples.

2.2. Analytical Techniques

2.2.1. Chemical Oxygen Demand

Filtered through a filtration paper (Ahlstrom, Espoo, Finland, grade 388), the COD of the samples was measured in triplicate using the spectrophotometric method for low concentrations (5–100 mg/L) following the Czech ISO standard ČSN ISO 15705.

2.2.2. UV-VIS Spectrometry

UV spectrometry is commonly applied in the water treatment industry to monitor organic matter. Specifically, absorbance at 254 nm is used as a proxy. However, with this parameter, mostly aromatic compounds or unsaturated bonds can be analyzed, which certainly does not cover all organic compounds. That is why the absorbance of the samples was measured (WTW photoLab® 7100 VIS, Xylem Analytics Germany Sales GmbH & Co. KG, WTW, Weilheim, Germany) at 5 wavelengths in the range of the UV-VIS spectrum (210 nm, 245 nm, 254 nm, 330 nm, and 615 nm) to cover a wider spectrum of molecules [30,31]. The wavelengths were selected considering that nitrates and nitrites absorb at 200–220 nm, conjugated dienes and unsaturated aldehydes and ketones at 220–250 nm, and generally, organic matter absorbs at 250–380 nm, and turbidity is observable at 380–750 nm [30]. Also, the wavelengths at which the subjected compounds exhibit their absorption maxima were assessed, specifically 239 nm (CIT), 276 nm (DCF), and 284 nm (CBZ) [29,32,33].

Absorbance was measured in a 10 mm cuvette in the samples filtered through a filtration paper (Ahlstrom, grade 388).

2.2.3. Liquid Chromatography

Ultra-high-performance liquid chromatography (HPLC) was used to detect the concentration of spiked substances (Agilent 1260 Infinity II, Agilent Technologies, Inc., Santa Clara, CA, USA). Combined chromatograms are attached in Supplementary Materials (Figure S1). The detection limit was 0.1 mg/L.

For purposes of the analysis, the samples were filtered through a 0.45 µm pore size syringe filter. When the experiment contained ClO₂, the oxidative reaction was stopped by adding an excess of the Na₂S₂O₃ solution (20 g/50 mL).

2.2.4. Fluorescence in Situ Hybridization

Fluorescence in situ hybridization (FISH) is a well-known molecular technique based on the utilization of fluorescent probes complementary to specific DNA sequences applied, for example, in the identification of the genera of bacteria present in activated sludge. To quantify probe-specific signals, a colorant such as DAPI (4′,6-diamidine-2′-phenylindole dihydrochloride) binds to all DNA present in the given sample [34].

Three analyses were performed (Olympus BX-51, Tokyo, Japan) using the procedure described in the FISH handbook by Nielsen et al. [35], utilizing the following probes in equimolar mixtures and, when appropriate, their competitors. For ammonia-oxidizing bacteria (AOB), the AOB mix consisted of Nso1225, Nso 190, NEU, NmV, and Cluster 192; for nitrite-oxidizing bacteria (NOB) Ntsp mix—an equimolar mixture of Ntsp662 and Ntsp712 and NIT3—was selected; and for polyphosphate-accumulating organisms (PAO), three probe mixtures were applied—PAO mix (PAO462, PAO651, and PAO846, targeting most of Candidatus Accumulibacter), PAOb mix (PAO462b and PAO846b, targeting Rhodocyclus tenuis), and RHC mix (RHC439 and RHC175a, targeting Rhodocyclus/Candidatus Accumulibacter and the majority of Rhodocyclaceae).

2.3. Coagulation Pretreatment

2.3.1. Coagulation Tests of Raw Water

After rough coagulation tests, a series of four tests in 800 mL vessels was conducted. Homogenized samples with four different doses of 4.1 wt.% Fe₂(SO₄)₃ (later, this solution is only mentioned as Fe₂(SO₄)₃) were mixed in a Flocculator Tester JLT4 (VELP Scientifica, Usmate, Italy) rapidly for 1 min (200 rpm), followed by 15 min of slow mixing (10 rpm). Then, the suspensions were allowed to settle. In supernatants, suspended solids (SS) and chemical oxygen demand (COD) were measured.

SS concentration was established by vacuum filtration through an ashless membrane filter (pore size 0.45 µm) of 30 or 50 mL according to Czech standard ČSN EN 872 [36]. The COD of samples filtered through a filtration paper (Ahlstrom, grade 388) was measured spectrophotometrically following the Czech standard for low concentrations (5–100 mg/L) ČSN ISO 15705 [37].

This experiment was performed only once, complying with common practice in wastewater treatment.

2.3.2. Coagulation Tests of Spiked Raw Water

Raw water samples (800 mL) were spiked with approx. 2 mg/L of each tested micropollutant separately and, then, with a combination of all three pollutants (approx. 2 mg/L of each). Subsequently, the optimal dose of a coagulant for raw water (0.6 mL of Fe₂(SO₄)₃/100 mL) was introduced, and a coagulation test was conducted, identically as already described above, in the case of raw water.

Samples were taken before and after the coagulation to perform COD and HPLC analyses as explained in previous paragraphs.

To the mixture spiked with all three analytes, different doses of a coagulant (0.5–0.65 mL of Fe₂(SO₄)₃/100 mL) were applied to assess how much the parameter of the micropollutant removal should be considered when determining the optimal coagulant dose.

2.4. Oxidation

2.4.1. Oxidation of Raw Effluent

The oxidation of COD present in raw effluent by ClO₂ was evaluated with different doses of ClO₂ within 5–10 min (4–20 mg/L) and 1 h after application (2–10 mg/L). Residual ClO₂ was measured using the CPR method.

2.4.2. Oxidation of Micropollutants in Effluent

Stand-Alone Oxidation

In 50 mL of effluent samples, 2 mg/L of each micropollutant was oxidized by ClO₂ with the initial concentration of 20 mg/L. The amount necessary to oxidize effluent COD was significantly exceeded, and maximum oxidation performance could be achieved.

Similarly, the combination of all three analytes (2 mg/L of each) in the effluent was oxidized by 5 and 10 mg/L of ClO₂ to evaluate whether any of the analytes are oxidized primarily.

In both situations, the residual concentration of ClO₂ and the initial and final concentrations of the analytes were determined.

Oxidation After Coagulation

After the coagulation described in Section 2.3.1, 100 mL of a supernatant was oxidized with ClO₂ (0–8 mg/L) to assess the change in the need for the oxidant due to coagulation. Residual ClO₂ was measured after 1 h of reaction.

Similarly, 100 mL of the spiked water supernatants from Section 2.3.2 was oxidized with an excess of ClO₂ (20 mg/L). After 1 h, residual ClO₂ was measured, and analytes from the samples before and after oxidation were determined.

2.5. Biological Activity on Sand Filter

2.5.1. Experimental Model Set-Up

The aim of the model design was to study the maximal ability of a sand biofilter to remove the selected micropollutants. Since it was expected that the stabilized biofilm would take a long time to develop and perform sufficiently, and the laboratory scale did not allow for the use of a flow-through model, the circular design of the model was prepared. A mineral carrier was used there, so that adsorption on the carrier particles did not appear, and biofiltration could be evaluated.

The WWTP effluent sample circulated from a 30 L bucket through a continuously aerated sand filter with an empty bed volume of 5 L. As the filling, waterworks sand was used with a particle size of 1–2 mm and a specific area of 2500–4000 m²/m³. The water flow rate through the filter was 4 L/h (HRT 1.3 h). The aeration was provided by a set of cylindrical aeration stones placed at the bottom of the filtration tank. The air pump aeration was reduced, so that sand flotation was prevented. In varying intervals (on average 14 days), fresh effluent samples were introduced to the system and replaced the circulating water. The entire model is depicted in Figure 1.

This way, the filter was inoculated, and a microbial culture grew on sand particles, creating a biofilm. The culture was later studied using FISH, as described earlier in the corresponding section (Section 2.2.4).

2.5.2. Biological Activity in Spiked Water

The water circulating through the sand biofilter was spiked with the tested micropollutants (CIT, CBZ, and DCF) dissolved in the effluent to obtain the total concentration in the system, targeting 4 mg/L. The observation of each analyte took at least two weeks before another analyte was added to the system.

2.5.3. Model Analyses

During the acclimatization period with only raw water, the samples were obtained once every 3–14 days into plastic bottles and stored at 4 °C. During the experiments with micropollutants, the samples were taken and stored in the same way with a frequency of 1–3 days.

The samples were filtered through a filtration paper or a syringe filter with subsequent analyses of COD, UV/VIS absorbance, and concentration of micropollutants, as already described in Section 2.2.

2.5.4. Biofilm Culture Analysis

First, 60 g of the sample from the reactor, consisting of sand, water, and biomass, was extracted from the model using a glass tube. The biomass was roughly separated from the big sand particles. The sample was first observed under a light microscope and then fixed and hybridized following the handbook for FISH analysis by Nielsen et al. [35].

3. Results and Discussion

3.1. Chlorine Dioxide

3.1.1. Synthesis

As described earlier, the method of synthesis was selected, as it is easy and safe to apply. To decrease the reaction rate, both substances (NaClO₂ and NaHSO₄, see Equation (1)) were first dissolved individually and then gradually mixed with a significant amount of water. Still, the expected concentration was not reached within the first 12 h of reaction. After 12 h in the fridge, the concentration was 1007 mg/L in comparison to the stoichiometric concentration of 3000 mg/L. After one week, the concentration reached 1577 mg/L. The actual concentration of the stock solution was calculated before every laboratory day using the equation listed in STN 75 7151 [26].

3.1.2. Analysis

As the absorbance of ClO₂ in demineralized water at 360 nm is a suggested method to determine ClO₂ concentration in STN 75 7151 [26], and, to the best of our knowledge, it does not carry any interferences, this method was chosen to be the source of initial data for the calibration of the CPR method.

Neither the calibration nor the calculation principle is provided by the mentioned Slovak standard, stating only that it is based on the decrease in CPR absorbance at 570 nm [26]. It was also not found in any other literature source, so the calibration was conducted as the dependency of ClO₂ concentration on the decrease in intensity with respect to the blank sample (no added ClO₂ or reduced by Na₂S₂O₃). The calibration curve (correlation coefficient R² = 0.9992) and equation were determined from a set of experiments in demineralized water to avoid losses of ClO₂ (Equation (2)).

$$\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n} \mathrm{o}\mathrm{f} {\mathrm{C}\mathrm{l}\mathrm{O}}_2\left[{\displaystyle \frac{\mathrm{m}\mathrm{g}}{\mathrm{L}}}\right]=51.163·\mathrm{d}\mathrm{e}\mathrm{c}\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{s}\mathrm{e} \mathrm{i}\mathrm{n} \mathrm{a}\mathrm{b}\mathrm{s}\mathrm{o}\mathrm{r}\mathrm{b}\mathrm{a}\mathrm{n}\mathrm{c}\mathrm{e} [\%]$$

(2)

3.2. Coagulation Pretreatment

3.2.1. Coagulation Tests of Raw Water

Coagulation tests were performed with the WWTP effluent sample. From the initial experiment, a set of the next four coagulation jar tests in larger volume (800 mL) was derived, ranging from 0.5 to 0.65 mL Fe₂(SO₄)₃ in 100 mL of the sample. The optimum was reached after the addition of 0.6 mL/100 mL of the sample (COD 12.3 mg/L, SS 6 mg/L, and pH 3.7). The optimum dose was then used in the experiments with spiked water.

3.2.2. Coagulation Tests of Spiked Raw Water

Stand-alone coagulation with 0.6 mL of Fe₂(SO₄)₃/100 mL of the spiked raw water led to the low removal of CBZ and CIT (up to 4.5 and 6.3%, respectively). On the contrary, the observed removal of DCF was at least 70.0%. Similar results were obtained for the mixtures of the micropollutants without observed interferences. The decrease in COD of 6–14 mg/L was comparable to the decrease seen in raw water (6–8 mg/L).

The removal efficiency of CBZ corresponds to the previous observations; however, in the case of DCF, it is significantly higher than reported. In cases of CBZ and CIT, this supports the statement that coagulation is not a self-sufficient method for micropollutant removal [16,38].

The experiments with mixtures containing different coagulant doses showed no significant effect on the removal of micropollutants, so the optimal dose can be set without considering the removal efficiency.

3.3. Oxidation

3.3.1. Stand-Alone Oxidation

To detect and measure residual concentration, the minimal concentration of ClO₂ in a raw effluent was 4 mg/L, if determined within the first ten minutes, and 8 mg/L, if determined after 1 h. For higher concentrations, residual ClO₂ exceeded the upper limit of quantification for the CPR method.

After the 1 h oxidation of spiked samples with 20 mg/L ClO₂, the removal values obtained were as follows: 100 ± 0% of DCF, 86.2 ± 2.2% of CIT, and 7.9 ± 2.0% of CBZ. In the mixture of these substances, no change in the removal efficiency was observed. For all experiments, residual ClO₂ exceeded the upper limit of quantification for the CPR method.

The removal results are consistent with the trend observed by Hey et al. [39]. The exact percentage could not be compared since the ratio between ClO₂ and micropollutants in our experiment was significantly smaller when compared to their results [39]. Generally, the recorded oxidation results align with the low oxidability of CBZ and better oxidability of CIT and DCF, regardless of the application of any other common oxidants, i.e., ClO₂, H₂O₂, or O₃ [17,39,40].

3.3.2. Oxidation After Coagulation

The supernatant after the coagulation of raw effluent was oxidized for 1 h with ClO₂ (0–8 mg/L), and residual ClO₂ was measured. The experiments showed that, at a dose of 4 mg/L, 0.89 mg/L of ClO₂ remained in the solution. At higher doses, the concentration exceeded the measurable limit of the CPR method. Therefore, it was assumed that 3.11 mg/L was necessary for the oxidation of the compounds present in the raw effluent.

In the experiments with a spiked effluent supernatant after coagulation with doses of 5 and 10 mg/L of ClO₂, the concentration of residual ClO₂ was seen to be higher than the quantification limit of the CPR method. Only in the cases of the samples containing CBZ with 5 mg/L did the residual concentration range from 0.0 to 0.1 mg/L. However, surprisingly enough, in the mixture of all tested compounds, residual ClO₂ also exceeded the limit of quantification.

The concentration of DCF after coagulation and oxidation decreased to 0 mg/L regardless of ClO₂. In the case of CBZ, the removal efficiency differed according to the dose of the oxidant, reaching 19.7% and 33.9% for 5 and 10 mg/L of ClO₂, respectively. Interestingly, when both coagulation and oxidation were applied to CIT, only a small amount of the compound (<0.25%) was removed. Figure 2 shows the removal by coagulation and oxidation combined, as well as the contribution of each step.

In all cases, oxidation significantly improved the removal efficiency. In comparison to the observations of stand-alone oxidation in the excess of ClO₂, for the combination of coagulation and oxidation, the efficiency of the oxidation step dramatically decreased for CIT (from 86.2% to 3.5%) but increased in the case of CBZ (from 7.9% to 18.0–32.3% ±depending on the dose). The combination of the two methods did not affect the efficiency of DCF removal, as it remained at 100%.

3.4. Biological Activity on Sand Filter

3.4.1. Biological Activity in Raw Water

At an early stage, the sand filter was exposed to the circulating WWTP effluent, with the expected capture of present microorganisms and formation of a biofilm on sand particles.

As a fast indicator of organic content, the absorbance at 254 nm was measured in the samples taken in differing intervals. The accumulation period data were gathered for four months, as shown in Figure 3. During this period, fresh effluent was introduced to the system.

Surprisingly, except for the initial period, new feeding water seemed to dilute the organic substances, even though the opposite observation was expected. Usually, a sharp decrease in absorbance was observed after water exchange, followed by a slow increase until the next feeding. This may indicate that the products of microbial activity absorb better at 254 nm than the substances originally present in the effluent. These could, for example, be parts of extracellular polymers or bacterial cells released into the effluent from the model when the biofilm was not strong enough and well-established [41].

Later, COD and absorbance at multiple wavelengths were determined in the circulating water. The results are provided in Figure 4. Apparently, COD and absorbance at 245, 254, and 330 nm correlate mutually. However, at 245 and 254 nm, the absorbance gives a higher value, and even minor changes in COD correspond to very clearly recognizable changes in the absorbance signal. In this sampling period, the value remained approximately constant with a slight increase within the last 4 days until the next feeding. There, a sharp increase appeared at first, as expected. Immediately after that, the values decreased to the original level.

Also, absorbance at 210 and 615 nm was measured; however, the signals were constant throughout the observation period and therefore irrelevant to the observation of biological activity and the potential removal of micropollutants.

3.4.2. Biological Activity in Spiked Water

To monitor their biodegradability, dissolved micropollutants were added one by one into the water circulating through the sand biofilter. The initial samples were acquired, and then several daily samples were also taken and analyzed via COD, direct UV absorption, and HPLC.

Figure 5 demonstrates an evident decrease in CIT concentration throughout the sampling period. Also, a clear correlation can be seen between the concentration of the substance and the absorption at the substance-specific wavelength. The first-order kinetic constant resulted in 0.057 day⁻¹.

In cases of CBZ and DCF, no significant changes were registered in pollutant concentration or absorption at specific wavelengths within 14 days of exposure (

Table 2. Observation of Micropollutant Concentrations and Absorbance at Specific Wavelengths during Biodegradation.

Biodegradation [days]	Carbamazepine		Diclofenac
	Concentration [mg/L]	Absorbance at 284 nm	Concentration [mg/L]	Absorbance at 276 nm
0	3.185	0.3758	1.417	0.4191
1	3.229	0.3748	1.074	0.3938
4	3.298	0.3769	1.414	0.3858
6	3.301	0.3758	1.171	0.3829
7	3.359	0.3557	1.219	0.4208
8	3.244	0.3470	1.267	0.3885
12	2.943	0.3800	1.321	0.4072
14	3.108	0.3625	1.215	0.3919
Average	3.208	0.3686	1.262	0.3988

).

In cases of CIT and CBZ, the results are in line with the review by Ianes et al. [12], where CBZ and CIT exhibited median biological removal rates of 4% and 39%, respectively. In the case of DCF, the minimal removal was noted in our study, while Ianes et al. [12] report 21% median biological removal. Owing to the observations made by Gajdoš et al. [17], considering just CIT, CBZ, and DCF, the best removal at a conventional WWTP, followed only by ultrafiltration, was reported for CIT, while in the cases of the other two substances (CBZ and DCF), the average removal in the same process was in negative values. This fact better corresponds to our measurements. The variety of biodegradation performance may be caused by different bacterial communities and other crucial factors.

Chemical Oxygen Demand

COD results for CIT were aligned with expected values. The addition of the micropollutant (3.3 mg/L) to the fresh effluent led to an increase in COD from 36 to 49 mg/L. Subsequently, a decrease to 30 mg/L was observed as both CIT and the organic content present in the effluent were metabolized.

In the case of CBZ, COD increased first from 30 mg/L in the blank sample to 42 mg/L. During the next 14 days, COD rose to 52 mg/L. No similar trend was observed for other parameters. This may have been caused by the release of different substances from the biofilm. These substances may be easily biodegradable, as none of the observed UV wavelengths showed comparable results.

The addition of DCF caused an increase in COD, growing from 27 to 48 mg/L. This value remained constant with just small fluctuations during the whole sampling period of 14 days.

UV Absorbance

Apart from the substance absorption maxima, absorbance was measured at 245, 254, and 330 nm. They were all found to behave similarly regardless of the micropollutant tested. However, all changes were most remarkable at 245 nm, which also provided the greatest values. On the contrary, the smallest values and the least visible changes were observed at 330 nm. Only the addition of CIT caused a more significant increase in the 245 nm signal than expected. This may be due to remarkably close values of the substance—maximal wavelength (239 nm) and 245 nm. These observations are in line with those made for raw effluent.

3.4.3. Biofilm Culture

Under the light microscope, the sample was found to consist of ca. 60–70% biomass and 30–40% inorganic content. In Figure 6, two rotifers can also be seen. They indicate a stable biocenosis with longer sludge residence time, which can be expected in the biofilm culture [42].

Biofilm stratification splits the biomass into oxic (active) and inactive biofilms. Also, within the biofilm active layer, fast-growing organotrophs typically dominate and occupy outer spheres, while slower-growing lithotrophs appear in deeper parts of the oxic layer [15]. FISH analysis showed the low representation of nitrifiers (AOB and NOB) and PAO in the biomass. It can thus be estimated that most bacteria in the biological culture are organotrophs, which is consistent with the typical biofilm stratification. In addition, PAO needs an alternation of oxic and anaerobic conditions. Therefore, deeply anoxic or anaerobic conditions must occur. Figure 7 demonstrates signals from hybridization with the AOB mix, RHC mix, and Ntsp mix, as these three probe mixtures exhibited the strongest signal out of the six tested.

4. Conclusions

This article assesses coagulation, oxidation by ClO₂, and biodegradation as methods to remove three micropollutants, namely citalopram (CIT), carbamazepine (CBZ), and diclofenac (DCF). These substances have been chosen from the list provided by the new EU Directive, and the particular methods have been selected as potentially cheaper, less demanding, and sustainable alternatives to the currently most preferred techniques.

Coagulation alone has been found insufficient for all the compounds tested, not exceeding an efficiency of above 80%. The best results were obtained for DCF (ca. 70%), while for CBZ and CIT, the efficiency was below 10%. Nevertheless, coagulation has been seen to reduce COD, resulting in the smaller consumption of chemicals and lower energy requirements in further steps.

Oxidation by ClO₂ performed significantly better, as DCF and CIT were removed with 85% efficiency or higher (100% and 86.2%, respectively). In contrast, the removal of CBZ remained below 10%.

When combined, the efficiency recorded for CBZ and CIT changed dramatically. While the removal of CBZ reached up to 33.9%, depending on the ClO₂ dose, no removal was effectively noticed when applied to CIT. This phenomenon could not be explained by any literature available and requires further research. In comparison, DCF removal efficiency remained at 100%.

Experiments on the biofilm on the sand bed support have confirmed that only CIT is potentially biodegradable, while the other two micropollutants cannot be removed biologically. However, the rate of degradation of CIT was rather slow. This may have been caused by the low concentration of biomass, as the sand biofilter was not fully stabilized, and the high concentration of CIT present in the spiked water in amounts greater by several orders of magnitude than detected in actual wastewater (units of mg/L vs. tens to hundreds of ng/L).

In sum, only CBZ could not be removed by any of the tested methods or their combination with the efficiency required in the revised EU Directive, although the removal efficiency on the whole WWTP was not assessed. Just the stand-alone oxidation by ClO₂ was able to remove over 80% of both remaining micropollutants.

Future research needs to examine lower and, therefore, more realistic concentrations of micropollutants, as their behavior may differ. In this study, higher concentrations were examined due to instrumental and procedural limits, which may have affected the results. Obviously, realistic concentrations should be examined before applying these methods to full-scale conditions. Also, the remaining nine compounds specified in the EU Directive should be evaluated similarly to obtain a complex picture of the suitability of those methods for complying with the required micropollutant removal efficiency. Concerning the economic aspect, this factor still has not been assessed satisfactorily, as the realistic concentrations of micropollutants have not been examined yet.

It should also be noted that the removal of the parent compounds by any transformation processes means the production of metabolites and derivatives, which may be even more harmful than the original compounds, which resemble the problematics of ozonation. Also, in the case of chlorine dioxide oxidation, problematic chlorite and chlorate may be formed. These products must be analyzed and assessed, as they could be discharged to the environment in notable quantities. When dangerous compounds are produced, another step, such as a GAC filter or a membrane, should be implemented for their capture. The results of these analyses may affect the suitability of a method in the full-scale application.

In either case, a significant potential lies in quaternary treatment. Not only could dangerous compounds and their metabolites be prevented from entering the environment, but also, the application of these methods can improve water management, promote water saving and sustainability, while being economically, energetically, and environmentally friendly.