A process involving highly sensitive activated carbon-based adsorbers as “police filters” to maintain BPA concentration in water below 2.5 ppb

1 Introduction

Bisphenol A (BPA) is one of the most widely used compounds in industrial processes for the production of polycarbonate plastics, epoxy resins and polyvinyl chloride (Ighalo et al., 2024; Massahi et al., 2025). This includes beverage and food containers, toys, medical and dental devices, and even thermal paper receipts until the end of 2019 (Ighalo et al., 2024; Umweltbundesamt, 2021). In 2011, global consumption of BPA was recorded at 5.5 million tons (Ighalo et al., 2024). The global market size of BPA is projected to grow from 8.72 million metric tons in 2025 to 11.98 million metric tons by 2030. This corresponds to a compound annual growth rate (CAGR) of 6.56% over the forecast period from 2025 to 2030 (Mordor Intelligence, 2025). BPA is classified as an endocrine-disrupting chemical (EDC) (Kropp et al., 2021).

EDCs are chemical substances of either anthropogenic or natural origin that can cause adverse health effects by interfering with the natural biochemical functions of hormones (Gao et al., 2025; Kropp et al., 2021; Umweltbundesamt, 2016). This can negatively affect reproduction, but may also have a range of other adverse health effects (Lind et al., 2025; Plachetka et al., 2025; Song et al., 2025; Umweltbundesamt, 2016). Due to its hormone-like effects, especially its ability to mimic estrogen, BPA disrupts the normal function of the endocrine system, potentially leading to a wide spectrum of health issues such as diabetes, cancer, obesity, thyroid dysfunction, high blood pressure, insulin resistance, and reduced fertility (Ighalo et al., 2024; Yazid et al., 2024). Recent studies have shown that BPA exhibits estrogenic activity at concentrations as low as 0.23 pg/mL, indicating its potential to exert long-term effects on human health (Yazid et al., 2024). Moreover, it has been identified in human urine and blood samples, indicating widespread exposure. (Wu et al., 2020). As a result, various regulatory measures have been implemented, with specific restrictions differing depending on the country.

Inefficient treatment processes in wastewater treatment plants are contributing to an accumulation of BPA in the environment. The transport of BPA into soil and water resources is further facilitated when BPA-containing products are exposed to high temperatures or acidic substances. Open landfills exposed to elevated temperatures are among the main sources of environmental contamination, as BPA can leach into groundwater. The concentration of BPA detected in environmental waters varies by country and region. For example, river samples from the United Kingdom revealed concentrations ranging from 0.037 to 0.462 μg/L; in the Mediterranean region, levels between 4.5 and 30 μg/L have been reported. In China, surface waters contained 0.253 μg/L of BPA; in Brazil, raw water samples showed concentrations ranging from 0.0037 to 0.194 μg/L; and in Italy, an average concentration of 0.459 μg/L was detected in 13 brands of bottled water and 10 tap water samples (Ighalo et al., 2024; Ologundudu et al., 2025). Beyond its presence in environmental waters, BPA has also been detected at concentrations as high as 0.1 g/L in drinking water and 17.2 mg/L in leachate from hazardous waste landfills (Yazid et al., 2024).

According to Guerra et al., the average removal efficiency of BPA in wastewater treatment plants ranged from 1% to 77%. A study conducted in Germany showed that BPA typically enters treatment plants at concentrations ranging from 0.33 to 910 μg/L. After wastewater treatment, BPA concentrations were reduced to levels between <0.01 and 0.65 μg/L. The study also included a comprehensive analysis of BPA concentrations in European surface waters and sediments over the period from 2010 to 2022. Only 46% of the evaluated studies met the predefined quality criteria, indicating significant reliability issues in publicly available data. This may also explain the pronounced variability in BPA concentrations across different regions and countries, as contradictory results were observed. Against this background, it becomes evident that, in addition to the effective removal of BPA, improving environmental monitoring and reporting practices is of central importance. (Swansborough et al., 2025). Regulatory authorities in Australia (Environment Protection and Heritage Council, the National Health and Medical Research Council, and the Natural Resource Management Ministerial Council) have established a drinking water limit of 200 μg/L for BPA. In contrast, the regulatory authority in the United States (Minnesota Department of Health) has set a more stringent limit of 20 μg/L (Ologundudu et al., 2025).

The European Food Safety Authority (EFSA) has recommended that the concentration of BPA in drinking water should not exceed 2.5 μg/L (Amtsblatt der Europäischen Union, 2020). The European Union has adopted a comprehensive ban on bisphenol A (BPA) in food contact materials, which came into force on 20 January 2025. The corresponding regulation prohibits the use of BPA, its salts in coatings, plastics and packaging that come into contact with food. Whilst this represents significant progress, it should be noted that various bisphenol analogues, such as BPS, BPF, BPAF, and BPB remain widespread and often exhibit toxicity and environmental persistence similar to BPA (Da et al., 2016; Trela-Kobędza and Ajduk, 2025). There is a serious regulatory gap regarding their use, and experts are calling for immediate action. It is important to highlight that in many developing countries, particularly in Africa, there are currently no clearly defined or enforceable regulatory frameworks addressing such compounds, even though the presence of BPA has already been documented in various environmental and food-related sources (Ighalo et al., 2024). Both the identification of suitable alternatives and the development of simple and efficient methods for the treatment of such wastewater must be prioritized (Ighalo et al., 2024; Yazid et al., 2024).

Currently, various approaches, including biological, physical, and chemical processes, are being employed to reduce BPA levels in water and soil. Among these, physical treatment methods such as adsorption and membrane filtration are frequently used for the removal of BPA from water and wastewater due to their simplicity and flexibility (Ighalo et al., 2024). Among these methods, adsorption has been the most widely used technique for BPA removal over the last 10 years, as it is easy to apply, economical, and associated with low generation of secondary waste products (Ighalo et al., 2024; Yazid et al., 2024). Various adsorbents are used in adsorption processes, including activated carbon, which is versatile, cost-effective, and considered highly promising for large-scale BPA removal (Ighalo et al., 2024). Moreover, it exhibits several advantageous properties, such as a large specific surface area (500–1,500 m²/g), high porosity, a variety of functional groups, as well as pronounced chemical stability and thermal resistance (Cheng and Wang, 2024; Yazid et al., 2024). Adsorption with activated carbon has also been recommended by the U.S. Environmental Protection Agency as one of the best available technologies for the removal of non-biodegradable and toxic organic compounds (Cheng and Wang, 2024). Filter systems utilizing activated carbon are currently under discussion and have already been implemented for the removal of EDCs. However, a major drawback lies in the high investment costs, which are primarily attributed to the use of activated carbon. These costs could potentially be reduced through the reactivation of activated carbon (Kropp et al., 2021). This process is also associated with significant expenses, which explains why most current research is concerned with finding alternatives to activated carbon (Cheng and Wang, 2024; Kropp et al., 2021). These include various membrane separation processes such as nanofiltration (NF), reverse osmosis (RO), and membrane distillation, as well as alternative methods such as ozonation, biological degradation, UV irradiation, and oxidation processes, including the Fenton process, as well as combinations of these processes (Khoo et al., 2022; Molkenthin et al., 2013; Ramos et al., 2021; Wang et al., 2020; Xu et al., 2025; Yüksel et al., 2013). Advanced electrochemical oxidation processes have also shown promising potential for the degradation of BPA (Cerrón-Calle et al., 2025; dos Santos and Brillas, 2026; Gaber et al., 2026).

The biological degradation of BPA by microorganisms plays a central role, because this pathway allows BPA to be metabolised relatively easily. Bacteria are primarily used, as they are the most extensively studied and can be widely found and isolated in various environmental settings, such as the biomass of wastewater treatment plants. Fungi and algae also possess the fundamental ability to degrade BPA. A major drawback is that not all microorganisms exhibit the same degradation efficiency, and biological degradation is highly dependent on operational parameters such as pH, temperature, and a continuous supply of oxygen. These factors make the process technically demanding and require intensive support. Although microbial transformation of BPA generally leads to intermediate products with lower toxicity or estrogenic activity, current data on the resulting metabolites remain insufficient.

Nonetheless, all currently available methods for BPA removal exhibit specific advantages and limitation. The development of efficient processes to adsorb BPA from water remains a technical challenge. The studies discussed above are interesting and often very promising but they were mostly carried out at BPA concentrations that exceed levels actually found in the environment. To fill this research gap, the present work focuses on the development of adsorbent materials as “police filters,” which are intended to work under relevant conditions, where BPA is present at trace concentrations (0–5 ppb). In the case of sudden contaminant intrusion, the purpose of the resulting process is to reduce and maintain the BPA concentration of drinking water below the limit of 2.5 ppb set by the EFSA. Beyond practical applicability, this work aims to present an approach that combines material development with process suitability by taking adsorption performance and material availability into account and focusing on a straight forward process that can be deployed by involving DMSO as a green solvent and well-established large-scale techniques. It should be regarded as an attempt to combine adsorber preparation with process design. The proposed activated carbon-based technology is environmentally friendly and scalable. The resulting materials were developed to exhibit substantial adsorption capacity in the trace concentration range and to be regenerable. They were utilized in a FBR that can be operated in a continuous way by means of tandem systems (Barbe et al., 2006). As recently, highlighted by Hammerstein et al. (2022), carbon dispersion has a major impact on the adsorption performance and considerable efforts were dedicated to the preparation of a stable activated carbon dispersion.

2 Materials and methods

2.1 Preparation of the activated carbon coated packing elements (CPEs)

Our coating methodology utilizes a solvent carefully selected for its ability to soften the surface of the packing elements without inducing complete dissolution. This softening process facilitates the adhesion of activated carbon particles, ensuring they are securely fixed to the surface of the packing element. Furthermore, we have selected dimethyl sulfoxide (DMSO) for its favorable environmental profile as a green solvent, thereby aligning this procedure with contemporary sustainability initiatives.

The packing elements used in this study are made of high impact polystyrene (HIPS). HIPS is a rubber-modified polystyrene distinguished by its enhanced mechanical performance (epsotech, 2025) Moreover, HIPS exhibits thermal stability up to 85 °C. The material’s softening point is around 95 °C and the melting point at 235 °C (Filamentworld, 2023). In addition, HIPS is considered safe for human health, which, given its extensive use, including in the food industry, renders it a suitable carrier material for a broad spectrum of applications (epsotech, 2025; Filamentworld, 2023). To enhance surface area, the packing element is designed with a grooved outer surface and an internal structure composed of eight fins. The total surface area is specified by the manufacturer as 823 mm². Unless specified otherwise, all chemicals employed in this study were sourced from Thermo Fisher Scientific, Waltham, MA, USA. For the coating of the packing elements, 16 g of activated coal (AquaSorb MP23UF, Jacobi Group, Kalmar, Sweden) was dispersed in 400 g of DMSO using a Schott bottle and an Ultraturrax (T 25 easy clean digital ULTRA-TURRAX®, IKA, Staufen, Germany) for 15 min at 10.000 RPM. After preparation of the dispersion, five hundred packing elements were weighed and added to the dispersion. The coating process was subsequently carried out in an incubator (INFORS HT Ecotron Stackable Benchtop Incubator Shaker with Flask Clamps; Bottmingen, Switzerland) for 30 min at room temperature (Laboratory is air-conditioned RT = 21 ± 1 °C) and 200 RPM. To remove excess dispersion from the packing elements, the coated materials were filtered, and the surplus dispersion was collected in the Schott bottle for reuse in subsequent coating steps. The CPEs were then briefly rinsed with deionized water to remove the solvent and any remaining dispersion. Finally, the elements were dried at RT until constant weight was achieved. The optical observation of the CPEs (Figure 2) confirmed the complete surface coverage with activated carbon. We estimated the mass of activated carbon deposited on the surface of the CPEs. For this purpose, we weighed 400 packing elements before and after coating. We measured a mass increase of 0.1585 g or 0.39 mg activated carbon per CPE. This indicates the presence of a very thin coating on the CPEs.

2.2 Characterisation of the activated carbon coated packing elements (CPEs)

Subsequently, a comprehensive characterization of the coating morphology and porosity was carried out using high-resolution scanning electron microscopy (SEM) and nitrogen adsorption measurements at 77 K, as described in Section 2.2.2. To complement the morphological characterization, abrasion tests were carried out using the PTF 20E double drum friability tester (PHARMA TEST Apparatus engineering AG, Hainburg, Germany) to evaluate the mechanical stability of the coating. In each test, 50 coated and uncoated packing elements were subjected to a standardized load at 250, 500, 1,000, 3,000 and 5,000 revolutions. Due to the absence of detectable material abrasion, the uncoated packing elements were tested only at 250, 500, and 1,000 revolutions. Abrasion-induced coating loss was quantified gravimetrically by determining the mass differential of coated packing elements before and after the Abrasiontest. To ensure reproducibility, all gravimetric determinations of abrasion-induced coating loss were carried out in duplicate.

2.2.1 Microporosity and pore size distribution

Nitrogen adsorption measurements at 77 K were conducted using an Autosorb-1 gas analyzer equipped with a Krypton upgrade (Quantachrome Corp., Boynton Beach, FL, USA). Prior to isotherm acquisition, the samples were degassed overnight at 200 °C. Surface area values were calculated using the Brunauer–Emmett–Teller (BET) method, and pore size distributions (PSD) were determined using three different methods. Initially, the coated filler particles were subjected to a macroscopic inspection to obtain a preliminary assessment of the coating’s uniformity and overall quality. Subsequently, SEM images were recorded using a Jeol JSM 7401F Field Emission Scanning Electron Microscope (JEOL USA Inc., Peabody, MA, USA) equipped with Gentle Beam mode.

2.2.2 Evaluation of adsorption behavior of the CPEs

Langmuir adsorption isotherms were recorded using LC-MS (LCMS-2020 Single Quadrupole, Shimadzu, Europe). For this purpose, 1 to 10 CPEs were placed in 150 mL of a 53.56 ppb BPA solution and incubated overnight at RT and 100 RPM. All measurements were performed in triplicate.

2.3 Adsorption behavior and mixing characteristics

2.3.1 Characterization of the FBR’s adsorption behavior

Adsorption experiments were carried out using a FBR, which was initially filled with 1.5 L deionized Water (see Figure 1). Subsequently, 250 CPE pieces were weighed and added to the FBR. The air supply was initiated at 0.1 bar (g) with a flow rate of 10 L/h, followed by the activation of the pumps (120U/DV peristaltic pump, Watson-Marlow GmbH, Rommerskirchen, Germany). The BPA solution was continuously fed at three different flow rates, which were defined as follows: 3, 1, and 0.5 L/h. Three different initial concentrations of BPA, 50, 25, and 10 ppb, were used in these experiments. At 30-min intervals, samples were collected and analyzed via LC-MS. The bed capacity at 2.5 ppb BPA was approx. 860 μg (calculated from Figure 7 for 250 CPEs, each coated with 0.39 mg activated carbon).

2.3.2 Comparison of the FBR’S hydrodynamics with the ideal continuous stirred-tank reactor (CSTR)

In order to study the effect of residence time on the adsorption performance of the CPEs, a residence time distribution was measured at different volumetric flow rates of 3, 1, and 0.5 L/h, corresponding to an average residence times of 30, 60, and 180 min, respectively. Volumetric flow rates were determined gravimetrically by measuring the mass of deionized water that had flowed through the system over the course of 1 hour. For this purpose, a pulse tracer experiment was carried out, in which 3 mL of a saturated sodium chloride (NaCl) solution was rapidly added into 1.5 L of deionized water containing 250 packing elements within the FBR. The displacement of the saturated NaCl solution was monitored using a GLMU 400 MP multi-parameter conductivity probe (Senseca Germany GmbH, Regenstauf, Germany). Data acquisition was performed using ProfiSignal software (Delphin Technology AG, Bergisch Gladbach, Germany). In order to evaluate the flow characteristics of the FBR and to compare its residence time distribution (RTD) with that of an ideal continuous stirred-tank reactor (CSTR), the dimensionless RTD function E(θ) was plotted as a function of the dimensionless residence time θ. The determination of E(θ) was preceded by the calculation of the RTD function E(t), expressed in units of min^-1. This was based on the evaluation of the tracer concentration cᵢ(t). For this purpose, the electrical conductivity κ measured at each time point t was normalized by the maximum conductivity of the electrolyte solution κ _*max*, and subsequently multiplied by the initial concentration c₀ of the BPA solution after the addition of the electrolyte. Equation 1 serves as the basis for deriving the calculation formula for cᵢ(t) (Emig and Klemm, 2017, 236–237, 240,). The experimental residence time distribution function E(t) _*exp* was then calculated based on the concentration profile ci(t), using the integral expression defined in Equation 1 (Emig and Klemm, 2017). It should be noted that Equation 1 is valid exclusively for pulse input tracer experiments (Emig and Klemm, 2017).

E t exp = c i t ∫ 0 ∞ c i t d t (1)

To determine the experimental normalized RTD function E(θ) _*exp*, the hydrodynamic mean residence time τ was first calculated according to Equation 3 and then multiplied by E(t) _*exp* in accordance with Equation 2 (Emig and Klemm, 2017, p. 241).

E θ exp = τ * E t exp (2)

To determine the ideal and normalized RTD function E(θ) _*CSTR*, the ideal RTD function E(t) _*CSTR* was calculated according to Equation 4 (Emig and Klemm, 2017, 234, 242).

τ = V V ˙ (3)

E t C S T R = 1 τ * e − t / τ (4)

The normalized ideal RTD function E(θ) _*CSTR* was finally obtained by multiplying τ with E(t) _*CSTR*, analogous to Equation 2. The dimensionless residence time θ was calculated as the ratio of the respective measurement time t to the computed residence time τ.

2.4 Regeneration of the CPEs

The regeneration of the CPEs was assessed with various solvents, including ethanol (EtOH), DMSO, aqueous sodium hydroxide (NaOH) at concentrations of 1 M and 0.1 M, as well as deionized water. For each experiment, 20 CPEs were immersed in 100 mL of a 100 ppm BPA solution in Schott bottles and incubated at 25 °C and 100 RPM. After 30 min, a sample was collected and the BPA-loaded packing elements were separated from the solution and rinsed with deionized water. Subsequently, the loaded packing elements were transferred into 100 mL of the respective regeneration solution and incubated again at 25 °C and 100 RPM for 30 min. The regenerated CPEs were separated from the regeneration solution and rinsed with deionized water. The regenerated CPEs were immersed in 100 mL of a fresh 100 ppm BPA solution and incubated (25 °C, 100 RPM). The regeneration process was then repeated following the same protocol.

2.5 Analytical methods

All analytical measurements were performed by the Chemical and Environmental Analytics Core Facility (ChEAF). For LC-MS analysis all solvents and additives were at least of LC-MS grade: acetonitrile (ACN; A/0638/17, Fisher Chemical, Loughborough, UK), methanol (MeOH; M/4062/17, Fisher Chemical, Geel, Belgium), BPA (Thermo Fisher Scientific, Waltham MA, USA), D16 - BPA (LGC Standards, Teddington, UK), Ultrapure Water (membraPure GmbH, Hennigsdorf, Germany) The chromatographic separation of was performed on a Nexera UHPLC coupled with a PDA detector (Shimadzu, Europe) and on a LC-MS (LCMS-2020 Single Quadrupole; Shimadzu, Europe). 40 μL of each sample were separated on an Orbit 100 C18 analytical column (4.5 × 50 mm, 5 µm (M/Z Analysentechnik, Mainz, Germany). Oven temperature was set to 50 °C, autosampler temperature to 15 °C at a flow rate of 0.8 mL/min. A binary gradient was applied using Ultrapure Water (A) and MeOH (B) as follows: 20% B at 0 min, 90% B at 2 min, 90% B at 3 min, 20% B at 3.1 min and 20% stop at 5 min. The PDA detector was equipped with a D2&W lamp with a wavelength range of 190–800 nm. A wavelength of 280 nm was selected to measure the samples. An LC-MS (LCMS-2020 Single Quadrupole, Shimadzu, Europe) equipped with a Atmospheric Pressure Chemical Ionization (APCI) source, operated in negative mode. The following source parameters were set: Detector voltage 1.5 kV, Interface temperature 400 °C, DL temperature 200 °C, Nebulizing Gas flow 3.5 L/min, Dry Gas flow 20 L/min, Heat Block 250 °C. MS data were obtained in selected ion monitoring mode (SIM-mode) (m/z 227, m/z 241) with a event time of 0.1 s. The calibration was carried out over a concentration range of 1–50 ppb. Sample preparation for both LC-MS analysis and UHPLC measurement using the Nexera system coupled with a PDA detector was carried out by adding 1,000 µL of sample, 200 µL of EtOH, and 20 µL of an internal standard (concentration: 20 μg/L) into a vial, followed by thorough homogenization. The calibration was performed repeatedly between sample measurements to ensure reproducible results, eliminate pronounced matrix effects, and verify the reliability of the analytical instruments.

Data evaluation and quantification was carried out using LabSolutions 5.97 (Shimadzu, Europe). For method validation following parameters were tested: accuracy, injection stability, limit of determination (LOD), limit of quantification (LOQ), linearity, precision and long-term stability. An eight-point calibration curve (1–50 ppb) was prepared in 1 mL deionized water and 200 µL ethanol and spiked with 20 µL of D16 - BPA in ethanol (final concentration 20 ppb). The resulting calibration curves (linear, 1/x weighting) demonstrated linearity with correlation coefficients ≥0.995 for BPA. LOD and LOQ were determined after six replicate measurements of the 1 ppb calibration standard on each of three separate days. LOD and LOQ were defined as threefold and tenfold the signal-to-noise ratio (S/N), respectively, resulting in values of 0.07 ppb ± 0.02 ppb for the LOD and 0.25 ppb ± 0.08 ppb for the LOQ.

For quality controls (QC), six replicates at concentrations of 1, 1.5, 2.5, 5, 10, and 45 ppb BPA were prepared analogously to the calibration standards. Accuracy and precision were evaluated over three separate days. Additionally, injection stability was assessed by performing ten consecutive measurements of a QC sample at 5 ppb. Within-run and between-run accuracy were within ±15%, and the relative standard deviation (RSD) was below 15% for BPA, indicating successful method validation.

Long-term stability at room temperature was assessed after 20, 242, and 377 days by comparing the peak areas of stored standard solutions with those of freshly prepared standards at each respective time point, using three replicates per time. Stability was expressed as the percentage of the peak area of the stored solution relative to the freshly prepared standard, resulting in values of 99.6%, 106%, and 101%, respectively, demonstrating that stable conditions were maintained throughout the study period. The matrix effect was not assessed as all samples were measured in the same matrix of deionized water.

3 Results and discussion

3.1 Characterisation of the activated carbon coated packing elements

Figure 2 shows the successful coating of the packing elements. From a macroscopic perspective, the applied activated carbon layer exhibits a homogeneous and uniform distribution across the surface of the packing element.

With mass losses <0.024% during abrasion resistance tests (see Figure 3), the results revealed the high durability of the coating against friction and abrasion forces. Visual inspection further confirmed the absence of any discernible damages. This study focuses primarily on the combination of the material and the process. It does not intend to demonstrate an adsorber material produced under rigorous quality standards that would ensure consistent batch-to-batch reproducibility. Nevertheless, we observed an acceptable level of reproducibility throughout the process. We added this information into the manuscript.

3.1.1 Pore size distribution

The nitrogen (77 K) adsorption isotherms in activated carbon sample is presented in Figure 4. The isotherm is of type IV, according to the IUPAC classification (Sing, 1984), characteristic of mesoporous adsorbents. However, at low relative pressures, a steep increase is observed due to the formation of a film of adsorbed molecules at the inner pore walls. More specific, a sharp uptake at very low relative pressures (P/P₀ ≲ 0.01–0.05) is evident. This is classic micropore filling, implying a high population of micropores (<2 nm) superimposed on the mesoporous network—i.e., a hierarchical micro–mesoporous texture. The modest, fairly narrow loop between adsorption (filled symbols) and desorption (open symbols) over roughly P/P₀ ≈ 0.4–0.9 is consistent with capillary condensation/evaporation in mesopores. The branches are nearly parallel and do not exhibit a pronounced “necking,” which is often associated with an H4-type loop (slit-like pores and significant microporosity). Minor offset of desorption above adsorption suggests some network effects (pore blocking or delayed condensation) rather than strong bottle-necking typical of H2 loops (Thomson, 1871).

The pore structural characteristics of the activated carbon studied sample are presented in Table 1. The total pore volume (TPV), the BET (Brunauer-Emmett-Teller) surface area (Stephen Brunauer/P. H. Emmett/Edward Teller, 1938) and the pore diameter (calculated by different methods) being 0.52 mL/g, 880 m²/g and ∼2.3 nm respectively (Table 1). The measured characterstics are comparable with values reported in the literature for BPA adsorber materials (Din et al., 2025) and the selected activated carbon appears to be appropriate for the purpose of this research.

TABLE 1. BET surface areas and pore characteristics of studied sample.

Total pore volume at 0.99.

Mean pore size d_mean as 4,000∙TPV/SBET.

Pore size was determined from the pore size distribution (PSD) using the Barrett–Joyner–Halenda (BJH) method based on the Kelvin equation of the N₂ desorption branch.

Pore size was determined from the pore size distribution (PSD) based on the FMS, equation (Favvas et al., 2011).

Micropore size was determined from the pore size distribution (PSD) based on the DA (Dubinin-Astakov) approach.

3.1.2 Material topology (SEM)

The SEM images of the coating reveals a heterogeneity, providing a multi-scale morphology. As it can be observed in Figure 5, irregularly shaped carbon aggregates exhibit a rough, flaky appearance with varied sizes. The dispersion of the activated carbon particles provide a satisfactory coating, something which is obvious from the gap free and very clear grey color of the CPE’s surface in Figure 2. The holistic exported “image” of the three SEM images is that the activated carbon coating on the packing elements’ surface unveils a distinctive carbon–polymer architecture characterized by irregular carbon particulates intimately interfaced with the polymer’s surface. This final obtained morphology suggests a balance between substrate adhesion and the adsorption functionality of the activated carbon material. Nevertheless, the formation of activated carbon aggregates can impair the material’s adsorption capacity by obstructing access to the adsorption sites.

3.1.3 Adsorption behavior at RT (21 °C)

To evaluate the adsorption behavior of the packing elements, the Langmuir adsorption model was applied (see Figure 6). This model assumes monomolecular adsorption, in which a monolayer forms on a homogeneous surface (Sadegh Mazloom et al., 2020). As expected from previous works (Din et al., 2025; Supong et al., 2019; Mourão et al., 2024), the corresponding Langmuir fitting curve was in good agreement with the measured data. At an equilibrium BPA concentration of 2.5 ppb (2.5 μg/L), the adsorption capacity was higher than 4,000 μg/m² (4 mg/m²) coating. This adsorption capacity is hardly comparable with values from the literature because of the lack of studies in this BPA concentration range. However, in the extreme case of heavily contaminated water with a BPA concentration of 5.0 ppb (5.0 μg/L), this threshold value of 4,000 μg/m² means that 1 m² of coating is capable of adsorbing enough BPA to reduce the concentration of at least 1.6 m³ of contaminated water from 5.0 to 2.5 ppb (EU/EFSA limit for BPA). These findings confirm the intended “police filter” function of the developed CPEs. Figure 7 additionally presents the mass-based adsorption capacity. At an equilibrium concentration of 2.5 ppb (0.0025 mg/L), the adsorption capacity exceeded 8,500 μg/g (8.5 mg/g) of activated carbon. As shown in Figures 6 and 7, the equilibrium adsorption capacity remains constant over a certain range of equilibrium concentrations, even though the equilibrium concentration changes significantly. We already observed this phenomenon during preliminary works. We dedicated numerous iterations to optimize the measure BPA adsorption equilibrium at such low concentrations. As reported in Section 2.5, we fully validated our analytical method and have confidence in it. An early saturation seems to be unlikely and we believe, as discussed in Section 3.1.2, that this behavior may rather result from a limited accessibility of the adsorption sites.

3.1.4 Comparison of mixing behavior in the FBR with the ideal continuous stirred-tank reactor (CSTR)

The residence time experiments, conducted at flow rates of 1 and 3 L/h respectively, are presented in Figure 8. It was observed that varying flow velocities had no significant effect on the RTD within the FBR. This assumption was further supported by comparison with the RTD-function of a CSTR, as the experimental values are in very good agreement with the corresponding model. The results suggest that the employed FBR enabled a homogeneous distribution of the CPEs, which is crucial for both efficient trace-level adsorption and reliable regeneration of the packing element.

3.1.5 Impact of operating conditions on hydrodynamics and adsorption behavior on the “police filter” function of the FBR

The adsorption experiments were carried out with 250 CPEs in a FBR (reactor volume was 1.5 L). In the following sections, all flow rates are reported in units of reactor volumes per hour (RV/h) to ensure consistency and comparability. For reference, the applied flow rates correspond to the following reactor volume exchange rates: 3 L/h = 2 RV/h, 1 L/h = 0.67 RV/h, and 0.5 L/h = 0.33 RV/h. Furthermore, one reactor volume (RV) is defined as the volume that completely passes through the FBR, thereby corresponding to a complete exchange of the reactor’s contents.

The main objective of this series of experiments was to evaluate the suitability of CPEs as “police filters” under real dynamic conditions. To demonstrate the effect of CPEs, we calculated the outlet concentration that would result from a continuous supply of contaminated water to the reactor without adsorption materials. For this purpose, we used the residence time sum function, as our system behaves similarly to a continuous stirred tank reactor (CSTR). The two first experiments (blue circles and black diamond in Figure 8) was carried by continuously feeding water contaminated with BPA (9.79 ppb and 8.25 – approx. 4–3.3 the BPA limit set by the EFSA) at 0.33 RV/h and 0.67 RV/h, respectively. In both cases, a clear adsorption effect of the CPEs was observed in comparison to calculated outlet concentration under zero adsorption condition (blue and black lines in Figure 8). After 6 h (Figure 9), the exit BPA concentration slowly increased to reach a 1.5–1.6 ppb and still far below the EFSA limit. It could be concluded that the FBR worked as intended by maintaining the BPA concentration within the legally permitted range in Europe.

A further experiment was performed at higher flow rate (2RH/h) and initial BPA concentration (11,19 ppb–approx. 4.5 times the limit set by the EFSA) in order to challenge the FBR in its “police filter” function. Under these conditions, the outlet BPA concentration exceeded the limit of 2.5 ppb after approximately 4.5 h (9 RV). It is worth mentioning that, in the view of European standards, a BPA concentration of 11.19 ppb represents a severe contamination as the average concentration in water ranges between 0.029 ppb and 0.036 ppb (Swansborough et al., 2025). These results confirm that in the event of sudden contamination at high flow rate, the FBR was able to deliver drinking water for many hours before breaking through. They also indicate that rapid adsorption takes place on the surface of the coating. In addition, the established limits in the United States (Minnesota Department of Health) and Australia (Environment Protection and Heritage Council, National Health and Medical Research Council, and Natural Resource Management Ministerial Council) are adhered to, as they are set at 20 ppb and 200 ppb, respectively.

To illustrate the threshold-undershoot, an additional representation of the United States threshold of 20 ppb was provided in Figure 11. Consequently, the CPEs developed could be used in Europe, but also in the USA and Australia. Subsequently, further experiments were conducted at higher starting concentrations. At initial BPA concentrations ranging from 20.70 to 26.09 ppb (Figure 10), the output concentration remains below the EFSA limit until 2.0 to 2.3 RV have passed the FBR. At moderate Flowrate (0.67 RV/h), breakthrough occurs after 4 h and after 1.5 h at a flow rate of 2 RV/h.

In a comparative experiment with an initial concentration of 42.80 ppb and a flow rate of 2 RV/h (Figure 11), the limit was exceeded after just 1 RV and 30 min, which clearly marks the operational limit of the FBR.

In Figure 11, the flow rate of 0.67 RV/h was investigated using two different initial concentrations to enable reliable comparability with the data obtained at 0.33 and 2 RV/h. As the experiments at 2 and 0.33 RV/h were conducted with starting concentrations of 42.80 ppb and 56.95 ppb, respectively, the measurements at 0.67 RV/h were performed using initial concentrations of 42.80 ppb and 53.02 ppb. This approach was chosen to ensure that any differences observed in the concentration profiles could be attributed exclusively to the flow rate, rather than to variations in the initial concentration. Nevertheless, even at very high trace concentrations, the United States regulatory limit of 20 ppb was undershot after 12 RV at a flow rate of 2 RV/h. This perspective demonstrate the considerable impact of local regulations and policies on the design and required performance of such fourth stage systems.

3.2 Regeneration experiments of the saturated packing elements

This study focuses primarily on the material’s police filter properties and its suitability for process applications. For the sake of completeness, we have also investigated the regeneration of the CPEs, albeit on a limited basis. Regeneration experiments using ethanol revealed significant structural degradation of the packing element. The material lost its mechanical integrity, became pliable, and was no longer able to retain its original form. Consequently, regeneration with EtOH was not further pursued. In contrast to ethanol, DMSO does not exhibit significant physical, environmental, or health hazards. DMSO has a higher boiling point of 189 °C and, with a lower explosion limit of 2.6 vol% and an upper limit of 42 vol%, features a considerably broader flammable range compared to Ethanol (Thermo Fisher Scientific, 2009). From a safety engineering perspective, this makes DMSO particularly advantageous for large-scale applications. As shown in Figure 12, DMSO proved to be the most suitable solvent for regeneration purposes among all solvents tested, as 75% and 60% of the original adsorption capacity were restored after the first and second cycles, respectively.

Desorption experiments with 0.1 M NaOH also demonstrated promising potential for restoring the adsorption capacity of the CPEs. After the first desorption cycle, the adsorption efficiency decreased to 45%, but was reduced by only an additional 6% following the second cycle, resulting in a final efficiency of 39%. In contrast, desorption attempts using water and 1 M NaOH were ineffective, yielding no appreciable recovery of adsorption performance. It was observed that after the first desorption experiment using water and 1 M NaOH, the remaining adsorption capacity of the CPEs was reduced to only 18%–19%. The slight increase of 2%–3% following the second desorption cycle is likely attributable to measurement uncertainties or natural fluctuations in the recorded values. Overall, the desorption experiments with water and 1 M NaOH did not yield satisfactory results. The residual performance of the CPEs after regeneration was insufficient, and thus both desorption agents cannot be considered suitable for further application. This lack of efficiency can likely be attributed to the low aqueous solubility of BPA, which is approximately 0.3 g/L (Merck, 2025).

4 Potential treatment steps for the resulting regeneration solution

Through the regeneration of the CPEs using DMSO, the BPA adsorbed onto the activated carbon layer is desorbed and enriched in the DMSO. To enable the reuse of the regeneration agent, however, it must be purified, and the concentrated BPA must be separated and concentrated. Various methods suitable for this purpose are illustrated in Figure 13.

A preferred method for the degradation of the remaining regenerated A is the photo-Fenton process (highlighted in red), which is discussed in more detail in the following sections. Since biological degradation also represents a possible option for the removal of regenerated A, it is briefly described as well. The aim is to integrate the resulting intermediates into industrial and economic processes, thereby establishing a fully closed-loop circular economy system.

To separate BPA enriched in the regeneration solvent (DMSO), the application of suitable membrane separation processes is required. Two processes are conceivable as potential purification steps, including nanofiltration (NF) and membrane distillation. Membrane technology is already used for the removal of micropollutants in water and wastewater (Khoo et al., 2022). Numerous review articles have described the application of various membrane separation processes for the elimination of BPA from different aqueous media (Ramos et al., 2021; Yüksel et al., 2013). Therefore, these technologies can be considered well-established, with a wide range of membranes having already been extensively investigated (Yüksel et al., 2013). Earlier studies demonstrated high retention rates of 83% to ≥98% when using polyamide NF and RO membranes, even at an initial BPA concentration of 100 μg/L (Kimura et al., 2004; Yüksel et al., 2013). Membrane distillation is also considered a promising alternative for the removal of micropollutants due to its versatile range of applications (Ramos et al., 2021). Unlike conventional membrane separation processes such as nanofiltration (NF) and reverse osmosis (RO), membrane distillation can be operated at low-medium operating temperatures and ambient pressure. In contrast, there is extensive literature on cleaning protocols for other membrane processes, such as NF and RO (Ramos et al., 2021). In view of the highly variable retention rates and the currently limited number of studies addressing membrane cleaning in the context of membrane distillation, the initial investigation of DMSO purification using NF and RO represents a methodologically sound and promising approach.

After successful purification of BPA-enriched dimethyl sulfoxide (DMSO), the subsequent investigations concentrate on the selective degradation of BPA. Any intermediates that may form will be analytically identified and characterized. Taking into account their properties and potential for value creation, they will be considered for further utilization in industrial and economic applications.

The Fenton process represents one possible method (Wang et al., 2020). It constitutes an advanced oxidation process (AOP), which can be coupled with UV radiation to form the so-called photo-Fenton process (Molkenthin et al., 2013; Wang et al., 2020). Using the Fenton reagent, BPA can be effectively degraded through oxidative processes. Under optimized reaction conditions, such as an appropriate concentration ratio of hydrogen peroxide (H₂O₂) to ferrous iron (Fe²⁺), an adjusted pH value and controlled reaction temperature, complete mineralization to carbon dioxide (CO₂) and water (H₂O) can be achieved (Wang et al., 2020). Furthermore, the introduction of a suitable light source, such as ultraviolet (UV) radiation, can markedly enhance the rate of the underlying oxidation reaction (Molkenthin et al., 2013). Another approach is the biological degradation of BPA by various microorganisms such as bacteria, fungi, or algae. BPA-degrading bacterial strains are widespread in the biomass of sewage treatment plants, among other places, and can be selectively isolated. This would be advantageous, as the coated packing elements used in this study are intended for use in such wastewater treatment systems. Biodegradation could thus serve as an existing downstream treatment step, with the purifying effect of the carrier materials being used upstream in the process. Although microbial degradation of BPA generally leads to intermediate products with lower toxicity or reduced estrogenic activity compared to the parent compound, it must be emphasized that the available data on the intermediates formed are still incomplete. This represents a potential risk factor, as some known degradation products may be even more toxic than BPA itself, which is why the Fenton process would be more advantageous for the time being. In addition, the effectiveness of biological BPA degradation depends heavily on the selection of suitable microorganisms (Zielińska et al., 2019). Furthermore, certain operating conditions such as pH, temperature, and oxygen availability must be maintained in all degradation processes to ensure optimal performance (Wang et al., 2020; Zielińska et al., 2019). Although the Photo-Fenton process has comparatively high energy consumption, which directly leads to increased operating costs, and requires considerable amounts of H₂O₂, its advantage lies in the complete mineralization to CO₂ and H₂O without the formation of significant intermediate products. Despite the limitations mentioned above, the degradation strategies presented provide a solid basis for the targeted removal of residual BPA, whereby the very low concentrations used in this study suggest a correspondingly low potential risk to human health.

5 Conclusion

In this work, CPEs intended to act as “police filters” in FBR were successfully prepared. They are characterized a mechanical stability, which was confirmed by a series of abrasion test. This particular function requires high adsorption kinetics and a considerable adsorption capacity a trace concentration. A Macroscopic analysis of the developed CPEs confirmed a homogeneous and uniform distribution of the activated carbon layer. The hierarchical micro-mesoporous texture was confirmed by both SEM and adsorption measurements using N2. The adsorption isotherms exhibited a significant uptake at low relative pressures, which is indicative of microporosity, and suggests enhanced accessibility through mesoporous pathways. This structural combination is particularly beneficial for adsorption processes, as it enables a high specific surface area alongside improved molecular diffusion and access to the microporous domains. The existence of this microporosity could explain the lack of BPA recovery using water or 1 M NaOH. Furthermore, SEM images of the activated carbon-coated polymer substrate reveal a heterogeneous, multiscale morphology. As expected from previous works carried out with activated carbon, the adsorption behavior of CPEs can be described by the Langmuir model and an adsorption capacity greater than 4,000 µg BPA/m² of coating was measured. In most studies, significantly higher initial concentrations of BPA were used compared to the present work. For example, Yüksel et al. investigated a BPA concentration of 50 mg/L (Yüksel et al., 2013), while Kimura et al. selected an initial concentration of 100 μg/L (Kimura et al., 2004). In another study, an even higher BPA stock solution with a concentration of 1,000 mg/L was used for the adsorption experiments. Both untreated and biochar-modified banana peels were employed as adsorbents.

The maximum BPA uptake was achieved at an initial BPA concentration of 240 mg/L. For the untreated material, adsorption equilibrium was reached after approximately 200 min, whereas for the biochar-treated banana peels, equilibrium was attained after 150 min.

The maximum adsorption capacity of the raw material was 87.9 mg/g after 200 min and remained constant thereafter. In comparison, the biochar-modified banana peels exhibited a significantly higher adsorption capacity of 131.5 mg/g after 150 min, which likewise remained unchanged thereafter. (Din et al., 2025). The cited study demonstrates the high adsorption performance of biochar; however, all experiments in the referenced studies were conducted under concentration conditions that differ from real environmental conditions and therefore only partially reflect actual environmental challenges. In contrast, the experiments conducted in this study explore the dynamic adsorption of BPA at concentrations ranging 8.25 from to 56.95 ppb (µg/L), thereby mimicking the operational conditions of a real case study. For this purpose, an analytical method established in this work allows for the detection and quantification of BPA at trace concentration levels. Adsorption experiments at different flow rates (in RV/h) in the FBR demonstrated effective removal of BPA in trace concentration ranges to maintain the EFSA limit of 2.5 ppb. The operation limit of the FBR was observed at initial BPA concentrations exceeding 42 ppb. Finally, several potential approaches were introduced for purifying the BPA inlet concentrations regeneration medium (DMSO) and for the selective degradation of residual BPA. Together, the experimental investigation of the CPEs regenerability, the combination of innovative materials with an environmentally compatible coating technology, and an application-oriented reactor design contribute significantly to the innovation of the presented process. To further assess and demonstrate the applicability of the proposed approach, future works including experimental studies involving real water samples (e.g. drinking water, tap water, river water, etc.) are currently being planned. Furthermore, the long-term stability of the coating and its regeneration for practical or large-scale operation remain uncertain and will be investigated in a dedicated long-term study at site in real waste water treatment plants.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Federica Raganati, National Research Council (CNR), Italy

Reviewed by: Alexsandro Jhones Dos Santos, São Paulo State University, Brazil

Uğur Salgin, Cumhuriyet University, Türkiye

Footnotes

BPA, Bisphenol A; BPAF, Bisphenol AF; BPB, Bisphenol B; BPF, Bisphenol F; BPS, Bisphenol S; CPE, Coated Packing Element; DMSO, Dimethyl sulfoxide; EDC, Endocrine-Disrupting Chemical; ED, Endocrine Disruptor; FBR, Fluidized Bed Reactor; m_ads, Total Adsorbed Mass, PDA detector; Photodiode Array Detector; RV, Reactor Volume; TPV, Total Pore Volume.