Removal of estrogen pollutants using biochar-pellet-supported nanoscale zero-valent iron

BET

Brunauer-Emmett-Teller

BP

biochar pellets

E1

estrone

E2

17β-estradiol

E3

estriol

EDCs

endocrine-disrupting chemicals

FTIR

Fourier transform infrared spectroscopy

HPLC

high-performance liquid chromatography

nZVI-BP

nano zero-valent iron-biochar pellets

SEM

scanning electron microscopy

In recent years, the fast development of industrialization caused the overusing of pesticides, fertilizers, antibiotics, and textile dyes (Yang et al. 2018). Because of their high toxicity, these contaminants have posed great risk to human health and ecosystems through bioaccumulation, antibiotic resistance, and various ecotoxicological effects (Blum et al. 2017; Yao et al. 2021). In the detection of endocrine disruptors (EDCs), estrogen (E1), 17β-estradiol (E2), and estriol (E3) are considered to have strong estrogenic activity (Liu et al. 2015). Although only an ng·L⁻¹ level is detected in aquatic environment, estrogen causes reproductive disorders in human and wild animals, such as feminization of male fish (Zhang & Zhou 2005; Sun et al. 2011). Therefore, the removal of estrogen pollutants has become a key issue in the world.

Various technologies are currently available for EDC removal within wastewater treatment processes, including biodegradation, adsorption, air stripping, and photolysis (Sim et al. 2011; Zeng et al. 2013; Wu et al. 2016). Adsorption is a common practice for estrogen removal from wastewater because of technological and cost advantages. A variety of adsorbents, including molecularly imprinted polymers (Le Noir et al. 2007), powdered activated carbon (Paune et al. 1998), carbon nanotubes (Joseph et al. 2011), and biochar (Ahmed et al. 2016), have been used for estrogen removal. Of these materials, biochar is regarded as one of the best adsorbents because of its high adsorption capacity, high specific surface area, and low cost (Chen et al. 2008; Zhou et al. 2016). However, biochar cannot be widely used because of its low strength and secondary pollution after loss.

nZVI, an emerging technology, was investigated extensively and showed high potential in treating organic contaminants in wastewater because of its large active surface area, high adsorption capacity, low cost, and environmental friendliness (Zhu et al. 2009; Fan et al. 2015; Dong et al. 2016). nZVI has been used to remove various organic pollutants (e.g., nitro-aromatic compounds (NACs) (Gu et al. 2015)), inorganic pollutants (e.g., heavy metal ions (Maamoun et al. 2021)) by adsorption. However, iron nanoparticles still have disadvantages, such as easily aggregated and oxidized in ambient conditions, which may decrease their reactivity and mobility (Huang et al. 2016; Li et al. 2016). Many methods are applied to overcome the agglomeration and passivation properties of nZVI such as immobilizing nZVI on solid material membranes (Wang et al. 2017c), coating the nZVI particles with carboxy methylated cellulose (CMC) (Lei et al. 2014), and doping NZVI with a noble metal (Wang et al. 2017b). BPs are an optimal choice because of their good adsorption performance, high compressive strength, and easy recovery.

Recently, there have been many studies on removing estrogens using biochars or nZVI. However, there have been far fewer studies on combining biochars and nZVI (Jarošová et al. 2015; Jiang et al. 2017a, 2017b; Wang et al. 2017c; Alizadeh et al. 2018). Accordingly, this study investigates the feasibility and mechanisms of the degradation of estrogens using nZVI supported by biochar pellets. The specific objectives of this study are (1) to investigate the effects of the solution pH, the dosage of nZVI on the biochar pellets, and initial concentration of estrogens on the removal of estrogens, and (2) to probe into the reaction mechanisms of removing estrogens using biochar-pellet-supported nZVI (nZVI-BP).

Analytic grade iron sulfate heptahydrate (FeSO₄·7H₂O), sodium borohydride (NaBH₄), and ethanol (99.7% v/v) were purchased from Sinopharm Chemical Reagent Co. (Shanghai, China). The estrogens, E1, E2, and E3, were obtained from Sigma-Aldrich (USA) with purities higher than 98%. The molecular structural formulas of the three estrogens are shown in Figure 1. HPLC grade methanol and acetonitrile were purchased from Aladdin Biochemical(Shanghai, China). The stock solutions of the target estrogens and the internal standards were individually prepared by dissolving each compound in methanol at a concentration of 1,000 mg·L⁻¹ and then storing them at 4°C in a refrigerator. Distilled water was purchased from Wahaha Company (China).

Preparation of the biochar: walnut shell biochar was produced by slowly pyrolyzing feedstock under nitrogen protection. The walnut shells were obtained from the Xinjiang Uygur Autonomous Region of China as a raw material and were milled into 1-mm powders using a grinder, washed with deionized water to remove residual sugar, and dried in an electric blast-drying oven. Then, the walnut shells were incubated in porcelain boats in an inner glass tube furnace microprocessor with a heating treatment. The final temperature was set to the desired value of 400°C, and the heating rate was set to 10°C·min⁻¹ (Tremblay 2018).

Preparation of the BPs: The walnut shell biochar, clay, NaHCO₃, NaSiO₃, and a certain amount of deionized water were placed in a beaker and stirred. The BPs were prepared in a pelletizer with a diameter of 1–1.5 cm. Then, the BPs were dried in an electric blast-drying oven at 45°C for 3 h, pyrolyzed in a muffle furnace with a heating rate of 10°C·min⁻¹ to 400°C, and maintained at that temperature for 3 h.

The synthesis process was conducted in a 500-mL three-necked, round-bottomed flask. The procedure is described as follows (1). First, 24.88 g of FeSO₄·7H₂O was dissolved in 200 mL of ethanol water (ethanol/water = 1/3, v/v) (2). Then, 50 mL of NaBH₄ reductant was added dropwise (2 mL·min⁻¹) to a solution containing BP and FeSO₄·7H₂O under vigorous magnetic stirring (3). The solution was shaken for another 30 min (Sun et al. 2006; Ghauch et al. 2009) after the addition of NaBH₄ was complete (4). The formed nZVI-BP was filtered and alternately rinsed with ethanol and degassed RO water three times each (Ghauch et al. 2009). In this synthesis, N₂ was bubbled into the solution during the entire process to maintain an inert atmosphere. The resulting samples were dried at 70°C in a vacuum overnight (Sun et al. 2006; Lv et al. 2011).

The BP and nZVI-BP were characterized using scanning electron microscopy (SEM), Fourier transform infrared (FTIR) spectroscopy, and the Brunauer–Emmett–Teller (BET) method. For the SEM test, a Quanta-250 environmental scanning electron microscope (FEI, Czech) was used. The samples were ground into powder and sprayed with gold for pretreatment. A NICOLET-380 FTIR spectrometer (USA) was used for the FTIR measurement using the KBr technique over the range from 4,000 to 400 cm⁻¹. The specific surface area and pore structure of BP and nZVI-BP were analyzed using a static nitrogen adsorption instrument at 77 K (Autosorb-iQ2-MP, USA).

The experiments were conducted in 150-mL conical flasks containing quantified estrogen solutions and adsorbents. In addition, 100 mg·L⁻¹ NaN₃ was added to inhibit microbial degradation. All adsorption experiments were performed in triplicate at 25°C with a rotation speed of 180 rpm. The initial pH of the solution was adjusted using hydrochloric acid or sodium hydroxide. To investigate the effect of pH on the estrogen adsorption, different pH values (5.0, 6.0, 7.0, 8.0, and 9.0) were selected in the study. The initial estrogen concentration was set to 1,000 μg·L⁻¹, and the dosage of nZVI-BP added into the estrogen solutions was 2 mg·mL⁻¹. To explore the performance of the adsorbent dosage on the estrogen adsorption, different dosages of nZVI-BP (1, 2, 3, 4, and 5 mg·mL⁻¹) were added to the estrogen solutions with an initial estrogen concentration of 1,000 μg·L⁻¹ and pH of 5.0. The estrogen adsorption kinetic experiments were performed at a BP/nZVI-BP dosage of 2 mg·mL⁻¹, an initial concentration of estrogen of 1,000 μg·L⁻¹, and pH of 5.0. The adsorbent was added to the estrogen solutions, and samples (1 mL) were taken to determine the relevant parameters at 10, 15, 45, 60, 120, 240, 480, and 720 min. To investigate the adsorption isotherm, BP/nZVI-BP (1 mg·mL⁻¹) was added to estrogen solutions with different initial concentrations (500, 1,000, 1,500, 2,000, and 2,500 μg·L⁻¹) while the pH was maintained at 5.0.

The estrogens were analyzed using an HPLC (Agilent-1100 Series, USA) composed of a Waters-C18 column (150 mm × 4.6 mm, 3.5 μm). The wavelength was set to 200 nm. The column temperature was maintained at 25°C, and the flow rate was 1 mL·min⁻¹. The initial mobile phase consisted of 70% water (A) and 30% acetonitrile (B). The level of solvent B was increased to 47% within 4 min, maintained for 8 min, and then returned to the initial settings within 3 min. The concentration of estrogens was determined using a standard curve with a correlation coefficient of 0.99.

Here, K_F/(mg·(g·L)⁻¹) and n represent the Freundlich constant and the heterogeneity factor, respectively,; C_e/(mg·L⁻¹) is the equilibrium concentration of the estrogen pollutants, and q_m/(mg·g⁻¹) and K_L/(L·mg⁻¹) indicate the maximum adsorption capacity and the Langmuir binding constant, respectively.

A comparison of the surface morphologies of BPs and nZVI-BP is shown in Figure 2. The results showed that the distinct and rich porosity of BPs provided a favorable possibility for nZVI to be trapped inside the pellets. The nonsupported nZVI was easy to agglomerate, and its reactivity was reduced, which was attributed to the small particle size, large specific surface area, and magnetic effect between the particles (Zhang et al. 2010a, 2010b, 2011a, 2011b; He et al. 2014). The nZVI attachment sites provided by BP prevented nZVI from agglomerating and improved the adsorption effects for the estrogens.

Figure 3 shows the FTIR spectra of BP and nZVI-BP. The peak in the range of 3,420–3,600 cm⁻¹ suggests the stretching vibration of −OH (Zhang et al. 2017). The peak at 1,613 cm⁻¹ indicated the stretching vibration of C = C or C = O (Shi et al. 2017). The adsorption of estrogen was improved by these oxygen-containing functional groups forming hydrogen bonds with atoms such as N and O in the molecular structures of the estrogens. The peak at 1,440 cm⁻¹ was due to the aromatic C = C stretching vibration (Zhang et al. 2017). The peaks at 1,081 and 797 cm⁻¹ indicate the stretching vibration of Si–O − Si (Keiluweit et al. 2010). In addition, there were large variations in the peak near 470 cm⁻¹ for nZVI-BP compared to the FTIR spectra of BP, which indicate the presence of nZVI (Devi & Saroha 2017).

Figure 4 shows the N₂ adsorption/desorption isotherms and the pore size distributions of BP and nZVI-BP. The hysteresis loop, as a typical characteristic, is observed in the N₂ adsorption/desorption isotherms of the two materials, indicating the existence of mesopores (Sun et al. 2013). In addition, the pore size distribution indicates the presence of micropores, which participated in the micromesoporous structure of BP and nZVI-BP. The detailed structural parameters are provided in Table 1. The BET surface area of nZVI-BP was 118.538 m²·g⁻¹, and the proportion of the micropore surface area was 68.1%, both of which were higher than those of BP (the BET surface area and the proportion of micropore surface area were 38.010 m²·g⁻¹ and 49.0%, respectively, for BP). Moreover, nZVI-BP had a total pore volume of 0.103 cm³·g⁻¹, and the proportion of the micropore volume was 38.8%, whereas BP had a total pore volume of 0.059 cm³·g⁻¹, and the proportion of the micropore volume was 15.3%. It has been reported that materials with large surface areas and narrow pore distributions have more sorption sites and high sorption energies, which significantly impact their adsorption behaviors (Pignatello & Xing 1996; Zhang et al. 2010a, 2010b).

Figure 5 shows the effect of time on the estrogen adsorption by BPs and nZVI-BP. It can be seen that the estrogen adsorption consisted of a rapid adsorption process, followed by a slow adsorption process until the adsorption equilibrium was reached. In the initial stage of the reaction, the rapid adsorption process was attributed to the large concentration difference between the material surface and the estrogen solution. As the reaction proceeded, the slow adsorption process occurred when the concentration difference decreased and the driving force weaken until the adsorption process finally reached equilibrium. The estrogen adsorption by nZVI-BP reached equilibrium after approximately 10 min, whereas the estrogen adsorption by BPs reached equilibrium after approximately 15 min. The adsorption effect of nZVI-BP on the estrogens was significantly higher than that of BPs because of its larger surface area. In addition, a primary cell could form spontaneously when nZVI-BP was added to the electrolyte solution, inhibiting the oxidation of nZVI (Dou et al. 2010; Lv et al. 2014) and improving the estrogen removal capacity. Note that the equilibrium adsorption capacity of the three estrogens varied as E2 > E1 > E3. This phenomenon was attributed to E2 having the highest water phase partition coefficient and the strongest hydrophobicity (Klimenko et al. 2002). Meanwhile, the keto group of E1 was more hydrophobic than the hydroxyl group of E3.

Figure 6 shows the estrogen adsorption effect of nZVI-BP at different pH values. It could be seen that the pH of the solution had a significant effect on the adsorption capacity of nZVI-BP. When the pH value was 5, the q_e values of the three estrogens reached a maximum, and the adsorption effects were best. As the pH value increased from 5 to 9, the adsorption capacity decreased significantly. This is likely because H⁺ promoted the corrosion of iron oxide on the surface of nZVI-BP and inhibited the passivation of nZVI under acidic conditions (Xie et al. 2017). Under alkaline conditions, iron oxides and passivation layers of Fe(OH)₃ can easily form on the surface of nZVI-BP, which likely hindered the contact between nZVI-BP and estrogen and the electron transport channel. In addition, the formation of a primary cell between BP and nZVI was stifled (Lai et al. 2012, 2013) when the solution pH exceeded 7.

The effects of the nZVI-BP dosage on estrogen adsorption were shown in Figure 7. It could be seen that the equilibrium adsorption capacity of the estrogens decreased significantly with increasing nZVI-BP dosage. However, the increase in the dosage may result in residual adsorption sites of nZVI-BP, and the superposition of adsorption sites led to a decrease in the adsorption area (Deng et al. 2011). Therefore, the adsorption capacity of estrogens per unit mass of nZVI-BP decreased. Conversely, with the increase in the dosage, the total surface area of nZVI-BP increased, leading to an increase in the removal rate of the estrogens.

Figure 8 illustrates the effects of different initial concentrations on the estrogen removal by nZVI-BP. With increasing initial estrogen concentration, the equilibrium adsorption capacity of nZVI-BP increased. The concentration difference between the nZVI-BP surface and the estrogen solution increased with increasing initial concentrations, which improved the adsorption rate of nZVI-BP. In addition, high initial concentrations increased the contact opportunities between the estrogens and the nZVI-BP, therefore improving the adsorption capacity.

Kinetic studies could investigate the equilibrium time and describe the removal processes, making such studies a significant method to determine the process mechanisms. The plots of q_e versus t for estrogen adsorption by BPs and nZVI-BP were shown in Figure 9, respectively. The fitting parameters were shown in Table 2. Comparing the values of the two models, the correlation coefficient (R² > 0.901–0.999) of the quasi-second-order dynamic equation was larger than that (R² > 0.843–0.978) of the quasi-first-order dynamic equation. Moreover, the theoretical equilibrium adsorption capacities of the three estrogens by BPs and nZVI-BP fitted to the quasi-second-order kinetic equation and were close to the actual equilibrium adsorption capacity, which indicated that the quasi-second-order kinetic equation could accurately reflect the adsorption kinetic law. In addition, the results indicated that chemical removal dominated the process (Blázquez et al. 2011; Zhou et al. 2017). Further, the K₂ value of E3 was lower than those of E1 and E2, indicating that the adsorption rate of E3 was the lowest.

The interaction mechanism between adsorbent and adsorbate could be inferred from the isothermal adsorption equation. Figure 10 demonstrated the adsorption isotherms of estrogens for BPs and nZVI-BP, respectively. In the Langmuir model, the adsorbent surface is uniform, and the adsorbate molecules form a monolayer around the adsorbent surface without interactions between the molecules (Jiang et al. 2017a, 2017b). However, the surface energies are heterogeneous and depend on the surface coverage in the Freundlich model (Lee et al. 2015). The fitting parameters were summarized in Table 3. The Freundlich isotherm were higher than those of the Langmuir isotherm by comparing the values of R² obtained by the two model, which indicated that the adsorption processes of estrogens by BPs and nZVI-BP could be better described by the Freundlich isotherm model and that the removal process was multilayer and heterogeneous (Foo & Hameed 2010). The slopes (1/n) in the study were below unity, reflecting a high affinity between the adsorbate and the adsorbent, and were indicative of chemisorption (Boparai et al. 2011). The K_F values were positively correlated with the adsorption capacity. The results showed that the removal capacity of estrogens by nZVI-BP is stronger than that by BPs.

In this study, nZVI-BP was successfully prepared and characterized as an efficient adsorbent of estrogens. We discussed the effects of the adsorption time, the solution pH, the dosage of nZVI-BP, and the initial concentration on the removal of estrogens by nZVI-BP. It was found that the dispersion and stability of nZVI particles on the support increased with decreasing agglomeration. The adsorption equilibrium of the estrogen removal by nZVI-BP was reached within 10 min. Further, the removal efficiency of estrogens by nZVI-BP decreased with increasing pH. The pseudo-second-order kinetic model accurately described the adsorption kinetics. The adsorption mechanism was found to be chemisorption, and the rate-limiting step was primarily surface adsorption. The Freundlich isotherm showed a better fit than the Langmuir isotherm. Therefore, the experimental results were consistent with the kinetic results.

This work was funded by the National Nature Science Foundation of China (No.31100938).

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