Optimization of the adsorption of diclofenac by activated carbon and the acidic regeneration of spent activated carbon

Micro-pollutants with a range of ng-μg/L and their metabolites in the aquatic environment, such as personal care products and pharmaceuticals, have unpredictable effects on the ecological system and human health (Zhuan & Wang 2020). Among pharmaceuticals, diclofenac sodium (DCF) is usually used as a non-steroidal anti-inflammatory drug. It has been classified as ecotoxic, hence it has recently been included in the list of primary pollutants by the EU (European Union) (Diaz-Angulo et al. 2020). DCF is a pollutant regularly found in aquatic environments, due to its high consumption rate (Sotelo et al. 2014). Micro-pollutants are generally not efficiently removed by conventional treatment processes due to their low biodegradability. DCF removal in sewage treatment plants is known to be only up to 21–40% (Zhang et al. 2008). Unfortunately, discharge standards are insufficient for monitoring such micro-pollutants (Kowalska et al. 2020). In order to increase the removal of such pollutants, conventional wastewater treatment plants should be improved with more effective technologies. The removal efficiencies obtained by some studied methods for removal of DCF are coagulation–membrane bioreactor (42±15.5%) (Park et al. 2018), combined ozonation–sonolysis (90–94%) (Fraiese et al. 2018), carbon-polymeric membrane (44%) (Nadour et al. 2019), biological degradation (10%) (Lee et al. 2019) and adsorption by activated carbon (above 90%) (Lee et al. 2019).

Among the methods used in wastewater treatment, the adsorption method has shown outstanding performance (Viotti et al. 2019). Adsorption is considered to be a flexible process in terms of both design and operation. Due to its excellent adsorbent properties, granular activated carbon (GAC) is used to separate the DCF. The use of carbon adsorbents obtained from renewable, low cost, abundant and natural products has been considered as a viable alternative. In DCF adsorption, adsorbents such as coconut shell (Purevsuren et al. 2016), Moringa aleifera pods (Viotti et al. 2019), cocoa shell (de Luna et al. 2017), olive stones (Larous & Meniai 2016) and Isabel grape bagasse (Antunes et al. 2012) were used. One limitation of adsorption by GAC is the depletion of adsorption pore sites. After the GAC saturated with the pollutant (i.e. spent), it can be replaced with fresh GAC, disposed of as solid waste, or regenerated for reuse (Huling et al. 2011; Hutson et al. 2012). Since adsorption is sometimes reversible, regenerating the adsorbent provides significant cost savings (Gil et al. 2019). The regeneration of GAC is a fundamental step in the sustainability of adsorption processes (Alvarez-Pugliese et al. 2019). Different methods for the regeneration of spent GAC have been developed, including thermal, chemical, microbiological and vacuum regeneration. Chemical and thermal regeneration techniques are commonly used for carbonaceous adsorbent regeneration. Chemical regeneration has emerged as an alternative to traditional thermal regeneration for reducing energy requirements and increasing regeneration efficiency (Salvador et al. 2015). Regeneration of GAC by chemical methods is an alternative method that can be performed in situ and on site (Hutson et al. 2012). Chemical regeneration can be defined as the desorption or decomposition of adsorbates using certain chemical reagents. In chemical regeneration, carbon attrition is not rapidly performed. In chemical methods, regeneration takes place through a chemical reaction or a pH change in the medium containing the GAC. Depending on the adsorbates are removed when the pH of the adsorbent surface changes. In chemical regeneration, mechanisms are extraction, pH changes, reaction/degradation and thermal desorption (Salvador et al. 2015). Traditionally, acid and alkaline chemicals are used for regeneration (Li et al. 2015).

Currently, there are few studies on GAC regeneration spent with pharmaceutical compounds. pH changes are often sufficient to displace the adsorption equilibria between the adsorbate and the adsorbent and promote desorption processes (Alvarez-Pugliese et al. 2019).

There is limited literature available on DCF adsorption by GAC and regeneration of DCF spent GAC. Therefore, this work investigated the DCF adsorption on coconut shell-based commercial GAC and regeneration of spent GAC. To determine the nature of the interaction between DCF and GAC, kinetic, isotherm and thermodynamic studies were performed. The adsorption process is optimized for high adsorptive capacity, while the regeneration process is optimized for high regenerative capacity at low operating cost. Optimization was carried out using the Taguchi Experimental Design. Optimization results of adsorption and regeneration will be used to evaluate the potential feasibility of processes.

Diclofenac sodium (C₁₄H₁₀C_l2NNaO₂) was supplied by Abdi İbrahim Co. Properties of Diclofenac sodium are given in Table 1. The coconut shell-based GAC used in the study (Tarkim-Carbon-200) was obtained from Tarkim-Carbon Co. in Turkey. The Brunauer-Emmett-Teller surface area, iodine content and dimensions of GAC are 1,100 m²/g, 100 mg/g, and 1.41–2 mm, respectively. Hydrochloric acid (HCl) was obtained from Merck with a purity of 37% (12.06 N).

The batch experiments were carried out with solutions prepared by dilution of the DCF stock solution. For the kinetic experiments, studies were conducted by shaking 1 g GAC with 50 mL DCF solution (40 mg/L) at different time intervals (1–60 min) with the initial pH of 6.0 and at 20 °C. For sorption isotherms, 1 g GAC in a 50 mL DCF solution with concentrations ranging from 10 to 100 mg/L was shaken for 30 min at 20 °C with an initial pH of 6. To evaluate the thermodynamic parameters for DCF adsorption, 50 mL of DCF solution (40 mg/L) with 1 g activated carbon was equilibrated at different temperatures (293–333 K). After DCF adsorption of raw GAC, an aliquot of the supernatant was filtered from an 0.45 μm membrane filter, and the residual diclofenac analyzed using a UV–VIS spectrophotometer (Hach-Lange DR 5000) at a wavelength of 281 nm.

To optimize the parameters affecting the adsorption of DCF on GAC and the regeneration of spent GAC, Taguchi Experimental Design was used. The parameters and their levels affecting the adsorption process were determined as pH (4.0–9.0), GAC amount (0.1–0.5), initial concentration (20–40 mg/L), temperature (20–80 °C), adsorption time (5–10 min), shaking speed (0.666–1.000 Hz) and GAC size (size 1 is <1.40 mm and size 2 is 1.40–2.00 mm). The response parameter for the adsorption process was determined as DCF adsorption efficiency (%).

DCF saturation of fresh GAC was accomplished with a concentrated DCF solution. The adsorption process was continued until the GAC could not adsorb the DCF molecules from the solution and the DCF concentration reached equilibrium in the solution. The spent GAC was mixed with the co-solution and the adsorption capacity of the regenerated GAC was determined. For regeneration experiments, the DCF spent GAC and co-solution (containing regenerating agent) were transferred to a glass beaker and exposed to regeneration by shaking the mixture. In order to determine the regeneration efficiency, experiments were carried out by shaking the mixture of regenerated GAC and DCF solution after GAC was washed several times with distilled water.

The parameters affecting the regeneration process were determined as HCl dosage (0.5, 3, 10 mL), regeneration time (5, 15, 30 min) and, temperature (20, 40, 80 °C). The levels of all parameters were determined by preliminary studies. The response parameter for the regeneration process was determined as the adsorption capacity (mg DCF/g GAC) and cost of regeneration process (€). Since more than one response parameter was selected for the optimization of spent GAC regeneration, the regeneration process was optimized as multi-parameter optimization. In the analysis, the responses were accepted as equal in weight. Energy and chemical consumption were taken into account in order to calculate the operating cost. The unit energy cost in the energy consumption of the magnetic stirrer was assumed to be €0.074/kWh. The unit cost of HCl was considered as €163/L.

Surface functional groups on the GAC surface were analyzed with Perkin–Elmer Spectrum 100 FTIR, a Fourier transform infrared spectrometer (FTIR). The morphological and structural properties of the GACs were analyzed with a Philips XL30 FEG, which is a scanning electron microscope (SEM).

Pseudo-first/second-order, intraparticle diffusion and Elovich kinetic models were used to examine the mechanism of the DCF adsorption process and to fit the kinetic data. The pseudo-first-order model is suitable for describing physisorption between adsorbed and adsorbent and specifies reversible interactions. The pseudo-second-order kinetic implies that the adsorption process takes place through strong chemical interaction forces between the adsorbent and the adsorbed (Viotti et al. 2019). The pseudo-second-order kinetic indicates that the adsorption capacity is proportional to the number of active zones filled on the adsorbent (Zhang et al. 2011). The pseudo-first-order and pseudo-second-order kinetic models may not be sufficient to explain the diffusion mechanism and the rate determination stage for adsorption. This can be explained by the intraparticle diffusion model (Kumar 2014). The intraparticle diffusion model can be used to define sequential mass transfer stages and to estimate the limiting step and the control mechanism in the adsorption kinetics (Maheshwari et al. 2013). Furthermore, the Elovich model can be successfully used to define the adsorption kinetics of the ion exchange system, to show that the adsorption process is a chemical process, especially an ion exchange process. In the Elovich model, an increase in or a decrease in increases the reaction rate (Genç & Dogan 2015).

Adsorption isotherms are required to gain some information about the adsorption mechanism, the surface properties and the tendency of adsorbent towards pollutants (Ghaedi & Kokhdan 2012). Adsorption isotherms can comprehensively explain how the adsorbate interacts with the adsorbents for the understanding of the nature of the interaction. In this study, some isotherms have been used. The Freundlich isotherm assumes the heterogeneity of the surface, and the adsorption occurs in regions with different adsorption energy (Maheshwari et al. 2013). It gives the relationship between equilibrium liquid and solid phase capacity. According to the Freundlich theory, the Freundlich constant (K_F) is the relative parameter of adsorption capacity and an n value greater than 1 indicates favourable adsorption (Wang et al. 2007), where 1/n is a measure of adsorption density and surface heterogeneity, and its approach to 0 represents an increase in heterogeneity (Moradi et al. 2013). The n value within 1 < n < 10 indicates favourable adsorption and values outside the range indicates unfavourable adsorption (Liu & Wang 2013). The Langmuir adsorption model assumes that sorption is localized in a single layer and there is no interaction between the axes of the adsorbate molecules (Maheshwari et al. 2013). The Harkins–Jura isotherm model mainly explains multilayer adsorption and the presence of heterogeneous pore distribution on the adsorbent surface (Liu & Wang 2013). Halsey adsorption isotherm is suitable for multilayer adsorption, in particular, the fitting of the Halsey equation can be best used for heterotopous solids (Moradi et al. 2013). The isotherm and kinetic models used in this study are summarized in the Supplementary Material Table S1.

Adsorption kinetics provides extremely important information for understanding adsorption mechanisms (Viotti et al. 2019). Different types of kinetic models were examined for the adsorption of DCF on GAC. In kinetic studies, it can be said that DCF was in anionic form because the pH of the solution was about 6, which is above the pK value of DCF, which was 4.2. The kinetic rate constants and the linear regression correlation coefficient values for each model are shown in Table 2. From these results, it was observed that the linear plot for the pseudo-first-order kinetic model presents low correlation, showing a poor fit for the model; however, the linear plot for the pseudo-second-order kinetic model shows a better fitting with a high correlation coefficient value. The compatibility of DCF adsorption with pseudo-second kinetics indicates that it contains chemisorption as the rate-limiting mechanism, which occurs by electron sharing or exchange between DCF and GAC adsorption. In this study, the second-order kinetic constant k₂ and q_e were found to be 0.056 g/mg·min and 1.873 (mg/g), respectively. In the study conducted by Larous & Meniai (2016), they concluded that pseudo-second-order kinetics gives the best correlation for DCF adsorption with olive stone-based activated carbon. They found the k₂ and q_e kinetic constants as 1.5 g/mg⋅min and 3.10 mg/g, respectively. de Luna et al. (2017) also found that the pseudo-second-order kinetic is suitable for DCF adsorption with cocoa husk-based activated carbon. They found the k₂ and q_e kinetic constants as 0.39 g/mg⋅min and 5.67 mg/g, respectively. When adsorption occurs with chemical forces, the adsorbent and adsorbed substance form a chemical bond to share electrons (Mansour et al. 2018). It was determined that the q_e,cal value calculated by the second-order kinetic expression is quite lower than the q_e value obtained experimentally (q_e, exp: 5.005 mg DCF/g GAC) and that k₂ depends on operating conditions. Also, k₂ relates to how quickly equilibrium is reached (Sekulic et al. 2019). In this study, the k₂ value is relatively low; therefore, the time required to reach equilibrium is relatively long.

The Freundlich and Halsey isotherm models can be used for multilayer adsorption and heterogeneous surfaces with non-uniform distribution of adsorption heat. By comparing the R² values of the studied models from Table 2, it can be concluded that Freundlich and Halsey models fit better than others. In this study, the K_F coefficient for DCF was calculated as 0.243 (mg/g) (1/mg)ⁿ. This coefficient is related to the adsorbent (Sekulic et al. 2019). Lach & Szymonik (2020) reported that in adsorption of DCF on commercial activated carbons, the K_F coefficient ranges from 26.02 to 9.17 mg/g. It was observed that the activated carbon used in this study has a very low affinity for DCF. In addition, 1/n value was found to be 1.37. The surface inhomogeneity of the adsorbent is rather low when 1/n values are closer to 0 (Lach & Szymonik 2020). The values of the thermodynamic parameters for the adsorption of DCF on GAC are also presented in Table 2. The negative values of change in can be used to describe the randomness at the AC solution interface during the adsorption process (Kumar 2014). In addition, negative indicates that the more ordered arrangement of dissolved molecules is shaped on the adsorbent surface (Wang et al. 2007).

The adsorption of DCF on GAC is an endothermic process, which is suggested by the positive value of the (Moradi et al. 2013). Endothermic processes prefer higher temperatures across the external layer and within the pores of the adsorbent, which increases the diffusion rate. Also, values higher than 20 kJ/mol are characteristic of the chemical adsorption process (Viotti et al. 2019). In this study, the value was obtained as 19.14 kJ/mol. The absolute values of in the range of 10–20 kJ/mol indicate the π–π interaction, the van der Waals force and hydrogen the bonding, which corresponds to the physical adsorption process (Wang et al. 2007; El-Shafey et al. 2012). It can be said that the adsorption of DCF on GAC occurs with the effects of both physical and chemical forces.

The optimization study was carried out to determine the operating conditions that provide maximum DCF adsorption on GAC. All experimental data in the Taguchi Experimental Design and response parameter in terms of DCF adsorption efficiency (%) are shown in Table 3. The Taguchi Experimental Design should include operating conditions to ensure low, medium and high responses. If it was observed that any parameter had no effect on the yield in the preliminary experiments, it would not be included in the experimental design. Since the selected parameters had an effect on the responses in the preliminary experiments, the levels of these parameters were determined to have a low, medium and high effect on the responses.

The consistence of the model was confirmed by analysis of variance (ANOVA) with a F value of 91.99 and a P-value of less than 0.05. ANOVA results are shown in Table 4. A (pH) and F (shaking speed) parameters, which are not specified in the ANOVA table, were included in the error due to their insignificant effect on the response. Higher F values indicate that the models are significant. Model parameters with a P-value of less than 0.05 are considered to be significant. When the P-value is examined, it is seen that the model is significant for DCF adsorption. In addition, the parameters of GAC amount, temperature, adsorption time and GAC size were determined as important parameters affecting DCF adsorption.

In addition, Pareto analysis was performed in order to determine the effects of variables on the adsorption of DCF. Pareto analysis is a useful method of determining the significance of effects in order of magnitude and displaying the 5% significance threshold (t-value limit) and an additional threshold (Bonferroni limit) corrected for multiple tests. The t-values of the bars represent the square root of the F values on the ANOVA. The parameters with the t-value on the Bonferroni line are defined as strictly significant, the parameters with the t-value between the Bonferroni line and the t-value limit are defined as potentially significant, and the parameters below the t-value limit line are defined as insignificant (Asem et al. 2018; Abdulredha et al. 2019). As seen in Figure 1, the most effective parameter for DCF adsorption on GAC was the GAC size. Potentially significant parameters were GAC amount, adsorption time and temperature. The initial concentration, pH and shaking speed were found as ineffective parameters in the adsorption of DCF on GAC. The parameters indicated by orange in the Pareto graph (see online version for colours) affect the adsorption positively and the parameters indicated by blue affect the adsorption negatively. For example, an increase in GAC size negatively affects DCF adsorption, whereas an increase in GAC amount, adsorption time and temperature parameters positively affects the DCF adsorption.

Supplementary Figure S1 shows the determined optimum experimental conditions for the DCF adsorption. The optimum conditions were obtained as pH 4.0, GAC amount of 0.5 g, initial concentration of 20 mg/L, temperature of 80 °C, adsorption time of 20 min, shaking speed of 0.666 Hz and GAC size of 1 (<1.40 mm). At optimum conditions, 79.8% DCF adsorption efficiency was obtained, which is higher than from all the adsorption efficiencies in the Taguchi Experimental Design. DCF adsorption efficiency predicted by the model was 79.06. The result obtained with the optimization experiment is within the 95% confidence interval in terms of model estimation.

In this study, the regeneration process of spent GAC with HCl was also optimized by the Taguchi Experimental Design. Operating cost of regeneration process and adsorptive capacity of GAC after regeneration were determined as response parameters. Hence, optimum regeneration conditions that will provide the low experimental cost and high adsorptive capacity of the spent GAC were determined by multi-response Taguchi Experimental Design analysis. For determining optimum regeneration conditions, HCl dosage, temperature and reaction time were determined as the parameters affecting the responses. Optimum operating conditions were determined to provide maximum adsorptive capacity and minimum operating cost simultaneously. Details of the multi-response Taguchi Experimental Design analysis are presented in Supplementary Materials.

The L9 Taguchi Experimental Design with experimental parameters, their levels and the obtained response values for the regeneration of spent GAC are shown in Table 5. The adsorptive capacity of raw GAC was found to be 5.005 mg DCF/g GAC.

ANOVA analysis was carried out to determine statistically important control parameters. Table 6 shows the results of ANOVA from the calculated MRSN values. In ANOVA, F and P (%) are useful for quantitative evaluation of parameter effects. Also, the factor with a high percentage contribution value has a greater effect on the response variable. According to ANOVA results, the amount of HCl is the most contributing parameter to the responses and has the greatest influence of 76.59%. Although the parameter of time is included in the error due to its very low contribution, the parameter of temperature has an influence of 5%.

The parameters effects on multi-response signal/noise (S/N) ratio (MRSN) are shown in Supplementary Figure S2. Overall mean value was calculated as 4.40. As shown in the Supplementary Figure S2., the optimum operating conditions for spent GAC regeneration are the second level of the A parameter, the first level of the B parameter and the second level of the C parameter. These conditions correspond to the HCl dose of 3 mL, a reaction time of 5 min and a temperature of 40 °C. At optimized conditions, 6.847 mg DCF/g GAC adsorption capacity was found with a experimental cost of €0.051. Total MRSN value of 7.26 was found with optimized conditions. The MRSN value predicted by the model was 7.28, while the confidence interval was found as ±3.52. Therefore, it can be said that the MRSN value obtained as a result of the optimum experiment remains within the confidence intervals.

Depending on the properties of the adsorbate, regeneration takes place via a chemical reaction or a pH change in the medium containing GAC (Salvador et al. 2015). Acidic treatment of the GAC can affect adsorptive capacity through physical processes, including the breakdown of carbon surfaces, widening of micropores and a decline in GAC surface area (Karanfil & Kilduff 1999). The frequency of regeneration of GAC and its performance after each regeneration is very important for cost savings. Figure 2 shows the adsorption capacities as a function of regeneration cycle. During the first three regeneration cycles, the observed adsorbability of DCF on regenerated GAC was higher than raw GAC. This could be why the regeneration of hydrogen bond increased DCF adsorption after HCl regeneration. During the third to ninth regeneration cycles, there was no significant reduction in sorption. After 16 regeneration cycles, sorption was reduced by 47.1%. These results show that by HCl regeneration, the sorptive capacity of the GAC was protected and the useful life of the GAC was increased.

Raw, spent and regenerated GAC were characterized by SEM and FTIR spectroscopy. Figure 3 shows the SEM image of the raw, spent and regenerated GAC. The raw coconut shell-based GAC has a rough surface with small pores and cracks. Some salt particles appear to be scattered on the GAC surface due to the remaining phosphate or other metal compounds on the GAC after impregnation at the GAC production (Anisuzzaman et al. 2015). The DCF remains are observed on the SEM images of spent GAC. As seen in SEM images after regeneration, DCF residues on the adsorbent decreased.

Functional groups on the GAC surface were examined by FTIR spectroscopy. Figure 4 shows the various functional groups on raw, spent and regenerated activated carbons. Commercial activated carbons generally have lower structural complexity (Viotti et al. 2019).

When the FTIR spectrum of raw, spent and regenerated GAC were examined, it was observed that there were two distinct differences. The first is the presence of the peaks, which characterize DCF in the range of about 700–800 cm⁻¹ in spent and regenerated GAC (Biorad Laboratories, n.d.). The peaks seen below 1,000 cm⁻¹ belongs to the π (C–H) peaks. It can be said that the peaks in this region were formed by the adsorption of DCF by π–π interaction. These peaks cannot be removed by regeneration. The second important point observed with the FTIR spectrum is the change in the intensity of the peaks seen at 1,035 cm⁻¹ (raw GAC), 1,072 cm⁻¹ (regenerated GAC) and 1,082.6 cm⁻¹ (spent GAC). The intensity of the peaks was ranked as regenerated GAC> raw GAC> spent GAC. The peaks at this range are probably associated with phenolic CO and OH, and primer alcohol–OH. The peaks in this region are caused by adsorption and desorption of DCF by electrostatic interaction.

The possible functional groups on the surface of raw, spent and regenerated GAC are presented in Table 7.

Possible DCF adsorption mechanism on carbon-based adsorbent may include: (1) H-bonding between the surface oxygen groups on carbon and DCF molecules, (2) π–π interactions between the surface carbon (π–electron donor) and aromatic rings of DCF molecules (π–electron acceptors), (3) hydrophobic interactions (El Naga et al. 2019). When pH < 4.15, DCF species existed mainly in the form of organic acid. H-bonding interactions should dominate the mechanism. When pH > 6.5, DCF species existed mainly in the form of anions with –COO⁻ group. Repulsion effect between both negatively charged DCF and GAC was expected (Wu et al. 2020). When the raw and spent GAC FTIR spectra are examined, it is seen that the C=C peak of the spent GAC shifts to 1,541 and 1,538 cm due to the π–π interaction in the adsorbent and adsorbate benzene ring (Jin et al. 2015). Figure 5 shows the possible adsorption mechanism of DCF on GAC.

In this study, the adsorption of DCF, which is one of the pharmaceuticals evaluated by the EU in the class of priority pollutants, with commercial coconut shell-based GAC and the regeneration of spent GAC were investigated. The study was carried out in four sections and the following results were obtained.

• The characterization of the nature of DCF adsorption by activated carbon was carried out. According to the ionic form of DCF in the aquatic solution, both physical and mainly chemical forces are effective in the adsorption DCF on GAC. The adsorption process of DCF on GAC was found to be compatible with the Freundlich and Halsey isotherm, which characterizes the heterogeneous surface structure and endothermic nature.

• The DCF adsorption process has been optimized by the Taguchi Experimental Design. The optimum conditions obtained were pH 4.0, 0.5 g of GAC, initial concentration of 20 mg/L, temperature of 80 °C, adsorption time of 20 min, shaking speed of 0.666 Hz and GAC size 1 (<1.40 mm). As a result of these verification experiment performed with these optimum conditions, DCF adsorption efficiency of 79.8% was obtained and this value is compatible with the model predicted value of 79.06% at 95% confidence level. All parameters effective in adsorption were optimized for maximum DCF adsorption efficiency. Based on the Pareto analysis and P-values, it was determined that the most effective parameter in the adsorption of DCF on GAC was GAC size. The parameters of GAC dosage, adsorption time and temperature parameters were also determined to have a significant effect. During the adsorption process, π–π interaction, H-bonding and electrostatic repulsion interactions play an important role.

• Regeneration of DCF spent GAC with HCl was optimized for maximum adsorptive capacity at minimum cost simultaneously. The most effective parameter affecting the regeneration process was found to be HCl dosage with 75% effect. The optimum conditions were obtained as an HCl dose of 3 mL, the reaction time of 5 min and a temperature of 40 °C for the regeneration of spent GAC. As a result of the verification experiment carried out with the optimum conditions, adsorption capacity of 6.847 mg DCF/g GAC, experimental cost of €0.051 and MRSN of 7.26 was achieved. The MRSN value found is compatible with the model predicted value of 7.28 at 95% confidence level. When the effect of the regeneration cycle on the adsorptive capacity of the spent GAC was examined, it was found that at the end of a total of 16 regeneration cycles, there was a 47% reduction compared to the adsorptive capacity of raw GAC, which is 5.005 mg DCF/g GAC.

• DCF, an important microcontaminant, was removed with a high efficiency from water with a coconut shell-based GAC. Regeneration of depleted GAC with HCl increased the lifetime of GAC as well as adsorption capacity.

All relevant data are included in the paper or its Supplementary Information.