Facile fabrication of sea buckthorn biocarbon (SB)@α-Fe₂O₃ composite catalysts and their applications for adsorptive removal of doxycycline wastewater through a cohesive heterogeneous Fenton-like regeneration

Abstract

A facile, low-cost and novel route for successful synthesis of SB@α-Fe₂O₃ composite catalysts was described through a simple thermal conversion process from SB@β-FeOOH precursor. The products were characterized by X-ray diffraction (XRD), scanning electron microscopy (SEM), energy dispersive spectrometry (EDS) and Fourier transformed infrared spectroscopy (FTIR) analysis, respectively. The experimental results indicated that α-Fe₂O₃ phase nanoparticles in the SB@α-Fe₂O₃ composites appeared in spindly nanorods and were highly dispersed on the surface of sea buckthorn biocarbon (SB) platform. The mechanism of thermal conversion from SB@β-FeOOH precursor to SB@α-Fe₂O₃ products was discussed. Subsequently, the naked SB and resultant SB@α-Fe₂O₃ composites were applied in fixed-bed columns for the effective pre-adsorption of doxycycline (DO) from an aqueous solution. Breakthrough curves for DO adsorption were carried out at different empty bed contact times (EBCTs). The results showed that the adsorption capacity of SB@α-Fe₂O₃ was much better than that of SB at all tested EBCTs. The DO-saturated SB@α-Fe₂O₃ bed was almost completely regenerated in situ through triggering a succedent heterogeneous Fenton-like regeneration process with an H₂O₂ dose of 1% w/v for 3 h, while the naked SB was hardly regenerated by a parallel processes. In view of the facile synthetic method of composites, the superior adsorption performance and the effective way of regeneration for consecutive pre-adsorption/regeneration cycles, the heterogeneous Fenton-like oxidation of DO-exhausted SB@α-Fe₂O₃ presents to be a promising and viable strategy for the regeneration of adsorbent. Regeneration mechanism of SB@α-Fe₂O₃ was in-depth investigated.

---

Introduction

Compared with the traditional chemical and biological strategies for the treatment of toxic and carcinogenic contaminated water, the adsorptive removal of organic pollutants from aqueous solutions is recognized as the most suitable approach due to its advantages of easy operation, low cost, nontoxic properties and high efficiency.¹ To date, numerous adsorbents, such as activated carbons,² clays,³ graphene oxide,⁴ zeolites,⁵ etc. have been exploited extensively and applied widely. Especially, the utilization of low-value, renewable, and environmentally friendly adsorbent is getting more and more popular. Supported by the sustainable utilization of natural carbon resources and the respect for the environment, the production of activated biocarbon adsorbents from agricultural waste residues has attracted significant attentions. Typically, classical agricultural wastes including rice husk,⁶ cocoa shell,⁷ cotton stalk,⁸ peanut shell⁹ etc., have been extensively utilized for the preparation of activated biocarbon adsorbents to deal with various organic pollutant like dyes, pesticides, antibiotics, heavy metal etc. In this regard, further extension of alternative high performance activated biocarbon adsorbents with a high adsorption capacity, substantial mechanical strength and low ash content is still in urgent need. Sea buckthorn, a branched and thorny deciduous shrub, is widely planted in many countries such as China, India, Myanmar, Finland, Nepal, Pakistan, Germany, Romania, Britain, Russia and many other high altitude areas.¹⁰ In the past fifty years, the great far-ranging production of sea buckthorn during the afforestation movement in China has resulted in large quantities of sea buckthorn waste (discarded sea buckthorn stem, branches and roots). The conventional routes for dealing with those sea buckthorn wastes, such as direct incineration with fire, casual disposal of abandonment and burying underground have been determined that they are characterized by serious environmental pollution and resource waste. Therefore, the exploration of finding the effective use of abundant waste sea buckthorn branches biomass is of particular importance in contemporary society, because waste sea buckthorn branches still have various primitive advantages of complete biodegradation, incredible abundance, and clear renewable. As to the fundamental structure of waste sea buckthorn branch, the chief chemical constituent of waste sea buckthorn branch is cellulose, hemicellulose, and lignin.¹¹ The existence of high contents of carbon within the molecular structure of cellulose, hemicellulose, and lignin make the waste sea buckthorn branch biomass becomes a desired precursor material for the production of biocarbon. To be important, the presences of talent pores within the natural sea buckthorn branch permit that the resultful biocarbon materials would inherit ample porous structure with excellent adsorption ability.

In comparison with traditional slurry suspensions of iron oxide or homogeneous Fenton reaction system, the efficiency of the heterogeneous Fenton-like process for the treatment of toxic wastewater has been increased significantly by applying an enrichment or pre-concentration sorption method prior to the oxidation of the contaminants by the Fenton-like process on the immobilized iron oxide. For instance, the primitive adsorption features of yeast cells and the high-level Fenton-like catalytic properties of Fe₃O₄ nanoparticles have been integrated for the effective removal and oxidative destruction of methylene blue dye contaminants in water.¹² Also, the regeneration of the exhausted SB@β-FeOOH sorbent was successfully achieved by initiating the heterogeneous Fenton-like oxidation over the anchored β-FeOOH by flowing H₂O₂ solution.¹³ Moreover, the adsorption capacity of iron modified bentonite (Fe-MB) was maintained after five successive adsorption/heterogeneous Fenton-like regeneration cycles.¹⁴ In this way, the common low efficiency of the heterogeneous Fenton-like oxidization strategies for the treatment of effluent has been improved observably by offering a pre-concentration or enrichment sorption prior to the oxidation of the pollutants. In particularly, usually large amount of reagents (H₂O₂) and much longer contact time needed in the slurry suspensions of iron oxide have been refrained through pre-adsorption route.

α-Fe₂O₃ (hematite), as a rhombohedrally centered hexagonal structure of corundum-type with a close-packed oxygen lattice in which two-thirds of the octahedral sites are occupied by Fe(III) ions, is the most thermodynamically-stable iron oxide under ambient atmosphere conditions. Among many available iron oxides which have been evaluated for utilizing as a catalyst in Fenton-like system, α-Fe₂O₃ (hematite) has drawn significant interest because of its positive catalyst effect, excellent chemical stability, low cost, abundance and thermodynamically-stable structure.¹⁵ Consequently, the synchronous use of α-Fe₂O₃ catalysts and H₂O₂ has been recognized as a promising Fenton-like agentia to oxidize the low-biodegradable organic compounds. For example, Liu and coworkers evaluated catalytic degradation of methyl blue in wastewater using α-Fe₂O₃ as heterogeneous catalyst in accompany with H₂O₂.¹⁶ Zheng C. et al. obtained composites consisting from the crystalline phase (α-Fe₂O₃) dispersed in the carbon framework and heterogeneous Fenton-like degradation of methyl blue achieved 96% with the optimal conditions of H₂O₂.¹⁷ Ursachi I and Stancu A conducted the catalytic removal process of methylene blue dyes from wastewater by α-Fe₂O₃/MCM-41 nanocomposites and H₂O₂, with the approximate degradation efficiency of 95–97%.¹⁸

In previous investigation, we verified that anchoring β-FeOOH nanoparticles on the surface of polyporous sea buckthorn biocarbon scaffolds could be achieved via a simple low-temperature hydrothermal process. Stimulated by the excellent performances of α-Fe₂O₃, in the present study, the subsequent thermal conversion processes of SB@β-FeOOH were further carried out to fabricate the SB@α-Fe₂O₃ composite catalysts. The adsorption properties of the SB@α-Fe₂O₃ products were investigated for the removal of DO from aqueous solutions. The regeneration in situ of the saturated SB@α-Fe₂O₃ composites were implemented by triggering the heterogeneous Fenton-like oxidation over the immobilized iron (α-Fe₂O₃) with an aqueous H₂O₂ solution after DO enrichment. In comparison with our previous SB@β-FeOOH catalytic materials, the SB@α-Fe₂O₃ exhibits a better thermodynamically-stable and recycled usage performance due to the good corrosion resistance, excellent environmental compatibility and fascinating physicochemical properties of α-Fe₂O₃. Generally, the conversion of waste sea buckthorn branches to a value-added biocarbon scaffolds and disguising them as a heterogeneous Fenton-like catalyst carrier would take full advantage of the plentiful and almost cost-free biomass material for adsorption. More importantly, regeneration mechanism of SB@α-Fe₂O₃ was in-depth investigated in this paper.

Experimental

Materials

Analytical grade urea, ferric chloride hexahydrate, zinc chloride, hydrogen peroxide, t-butanol, p-benzoquinone and sodium azide were supplied by xi'an Chemical Agent Corp and used without further purification. Distilled water was used throughout the experimental procedures. Sea buckthorn branches, obtained from Qinghai Province, were initially impregnated with the activating agent ZnCl₂ and then pyrolyzed in a tubular furnace at the temperature of 773 K under N₂ atmosphere for 1 h. The resulting solid product (SB powder) was recovered by filtration and washed with distilled water.

Synthesis of SB@α-Fe₂O₃

The preparation of the SB@α-Fe₂O₃ from the SB powder was carried out through the following steps: (i) the SB powder was immersed into the prepared solution (0.05 M FeCl₃·6H₂O and 0.2 M urea) and then the mixture was placed in a Teflon-lined stainless steel autoclave; (ii) the autoclave was sealed and maintained for 6 h at 353 K. After the hydrothermal reaction process, the resulting products (SB@β-FeOOH) were filtered, washed with distilled water, and dried in vacuum at 333 K for 6 h; (iii) the SB@β-FeOOH composites were transferred to SB@α-Fe₂O₃ at tube furnace and heat-treated for 2 h at 773 K under N₂ atmosphere.

Characterizations of SB@α-Fe₂O₃

The crystallographic structures of samples were characterized by X-ray diffraction (XRD) using Cu Kα radiation in the region of 2θ from 5° to 70°. With the purpose of measuring the surface morphology of the samples, scanning electron microscopy (SEM) images were taken on a Hitachi S-2700 scanning electron microscope. The EDS, equipped with Hitachi S-2700 scanning electron microscope, was employed to evaluate the iron, carbon, and oxygen distributions in SB@α-Fe₂O₃. Fourier-transform infrared (FT-IR) spectroscopy measurements were performed on a Bio-Rad FTS135 spectrometer in the range 4000–400 cm⁻¹ using a KBr wafer technique.

The pH of zero point charges (pH_PZC) of the SB and SB@α-Fe₂O₃ composite was measured with the following procedures according to reported literature:¹⁹ firstly, 50 mL NaCl (0.01 mol L⁻¹) solution was placed into 250 mL glass bottle. The solution pH was adjusted to successive original values between 2.0 and 12.0, and 0.1 g samples were also added to the glass bottle. Secondly, the glass bottles were filled with N₂ to remove the effect of CO₂ on the pH change, and then sealed and shaken at 313.15 K. Thirdly, after a desired contact time of 48 h, the final pH of solution was determined. Finally, the ΔpH is defined by the difference between the final pH and the original pH (final pH-original pH). The ΔpH was plotted against the original pH. The pH_PZC of samples is the solution pH at which the curve crosses the line of ΔpH = 0.

Adsorption tests

The adsorption performance of SB and SB@α-Fe₂O₃ was operated in up-flow mode fixed-bed column made from glass tube with a length of 15 cm and an internal diameter of 0.6 cm. The effect of SB and SB@α-Fe₂O₃ bed depth 2.0 cm, the influent DO concentration 10 mg L⁻¹, the various EBCTs ranging from 0.19 to 0.57 min and free pH 6 on the removal of DO in the column was studied. Effluent DO solution from the fixed-bed was collected at appropriate time intervals and then detected by a Jenway 6405 UV-vis spectrophotometer at λ = 351 nm. The adsorption performance capability of SB@α-Fe₂O₃ and SB was determined by equilibrium uptake capacity Q_e as the following relations:where Q_e is the quantity of DO fed to the SB@α-Fe₂O₃ divided by mass of adsorber at total flow time (mg g⁻¹), F is the flow rate of adsorbate DO (mL min⁻¹), C₀ is incipient concentration of DO (mg L⁻¹), t is the bed total flow time (min). Besides, M is the mass of adsorbent SB@α-Fe₂O₃ in column (g).

In Situ regeneration of SB@α-Fe₂O₃

After pre-adsorption experiments, the regeneration of saturated adsorbent was investigated by transferring saturated SB@α-Fe₂O₃ to an aqueous solution of H₂O₂ (1% w/v) for 3 h, which triggered the heterogeneous Fenton-like reaction of the contaminants. Then the sorbent was centrifuged and washed with distilled water before the next loading procedure of DO solution. The regeneration cycle of SB@α-Fe₂O₃ was repeated (at EBCT 0.57 min) as explained above to assess the regeneration ability in situ and the lifetime of SB@α-Fe₂O₃ in the succedent adsorption experiments. The regeneration efficiency (η) was calculated by eqn (2).where x_r is the DO adsorption capacity of the regenerated adsorbents and x_f is the DO adsorption capacity of the fresh adsorbents. The adsorption capacity x was evaluated by eqn (3).

Results and discussion

Characterization of SB@α-Fe₂O₃ and formation processes

XRD patterns of the SB powder (trace a), SB@β-FeOOH samples (trace b) as well as the intermediate state at 673.15 K (trace c) and heat-treated SB@α-Fe₂O₃ products at 773.15 K (trace d) were presented in Fig. 1, respectively. As shown in Fig. 1(a) of SB trace, the broad and strong bread-like diffraction peak around 2θ = 25° indicates that the SB can be assigned to amorphous structure. XRD pattern of the SB@β-FeOOH in Fig. 1(b) shows that almost all the reflections can be readily indexed to a tetragonal β-FeOOH phase (JCPDS no. 34-1266), and no other characteristic peaks were detected for impurities, demonstrating the unique presence of β-FeOOH on SB surface. After thermal conversion of SB@β-FeOOH precursors at 637.15 K, it can be seen that full profile of β-FeOOH disappeared in trace c (intermediate state at 673.15 K), and a small peak of (104) belonged to α-Fe₂O₃ characteristics can be easily observed. In contrast, the intensity of (104) peaks increases sharply with the rise of the sintering temperature to 737.15 K in Fig. 1(d), and some other diffraction peaks also can be founded, all of which are in agreement with the theoretical data of hexagonal structure α-Fe₂O₃ (JCPDS no. 33-0664). The strong intensity in Fig. 1(d) suggests that a good crystallization procedure of α-Fe₂O₃ has been achieved. In other words, SB@α-Fe₂O₃ was able to be formed facilely above the temperature of 737.15 K.

Fig. 1 XRD pattern of (a) the original SB powder, (b) SB@β-FeOOH samples, (c) intermediate state and (d) SB@α-Fe₂O₃ products.

The formation of the α-Fe₂O₃ nanoparticles on the structure of SB@α-Fe₂O₃ composites was further explored by SEM analysis. Fig. 2(a) demonstrated the naked porous SB architecture, which offered highly accessible open channels with a few microns in diameter. These plentiful native pores in natural sea buckthorn branch could provide more active site for adsorption of target DO. Fig. 2(b) is the SEM image of intermediate SB@β-FeOOH samples. It is clearly observed in Fig. 2(b) that the surface of SB biocarbon powder was covered by many spindly nanoparticles of β-FeOOH. The size of spindle-like β-FeOOH nanorods were 60–100 nm in width and 300–400 nm in length. The SEM images of SB@α-Fe₂O₃ products were shown in Fig. 2(c), and obviously, the loading of many α-Fe₂O₃ nanoparticles resulted in a rough surface topographic features of SB@α-Fe₂O₃. Further, the lack of scattering particles around the composites implied a sturdy adhesion between the SB scaffold and the α-Fe₂O₃ nanoparticles. At higher magnification of SB@α-Fe₂O₃ species image in Fig. 2(d), many α-Fe₂O₃ nanoparticles pervaded densely on the surface of SB, but the luxuriant pores could still be observed clearly. Such porous structure, inheriting form their parent SB scaffolds, maintained high absorbability features after α-Fe₂O₃ modification process. Fig. 2(d) also displayed the spindle-like morphology of α-Fe₂O₃ which was the typical crystal habit of β-FeOOH. This is because the decomposition reaction of β-FeOOH to α-Fe₂O₃ takes place within a microcrystal without remarkable variation in shape and the interparticle sintering leads to heterogeneity in nanoparticle size, as pointed by Parida KM.²⁰ Up to this point, the successful and adequate loading of α-Fe₂O₃ nanoparticles onto the surface of SB biocarbon was confirmed.

Fig. 2 SEM images of (a) morphology of original SB powder, (b) SB@β-FeOOH samples, (c) the SB@α-Fe₂O₃ products, and (d) an enlarged image of SB@α-Fe₂O₃ under greater magnification.

The EDS analysis of SB@α-Fe₂O₃ product was submitted to investigate the distribution of elements in the SB@α-Fe₂O₃ composites in Fig. 3. The analysis of element content announced that C, O and Fe were the main constituents of SB@α-Fe₂O₃, indicating the purity of the SB@α-Fe₂O₃ composites. Thereinto, the precise content of Fe is estimated up to 27.72%. Compared with the classical impregnation method,^21–24 the present thermal decomposition route hereby has obviously provided a better choice to synthesize the immobilized composites comprising Fe₂O₃ nanoparticles and carbon-related materials with higher content of Fe₂O₃. Such improved dispersion of the α-Fe₂O₃ nanoparticles onto the SB surface inevitably generates available regeneration sites, which will consequently facilitate heterogeneous Fenton-like process.

Fig. 3 EDS analysis for C, O, and Fe elements of SB@α-Fe₂O₃ samples.

To further illustrate the SB@α-Fe₂O₃ formation procedures, the FTIR spectra of the powdered SB, β-FeOOH, SB@β-FeOOH, and SB@α-Fe₂O₃ composites were highlighted in Fig. 4 and the wave number assignments were listed in Table 1. In Fig. 4(a), the strong peak observed at 1640 cm⁻¹ was due to stretching vibrations of the skeletal C=C.²⁵ In the spectrum of Fig. 4 (b) and (c), the broad peak between 3200 and 3500 cm⁻¹ can be assigned to the O–H stretching vibrations of β-FeOOH, which might be obscured by a great quantity of adsorbed water.²⁶ While in Fig. 4(d) (SB@α-Fe₂O₃), the peaks of 3200 and 3500 cm⁻¹ cannot be observed any more, indicating the disappearance of adsorbed water during the thermal decomposition. Moreover, compared with the FTIR spectra of SB@β-FeOOH, there is no peak at 696 cm⁻¹ (d-OH band of β-FeOOH)²⁷ in the curve of the SB@α-Fe₂O₃, further confirming that the dehydroxylation of the β-FeOOH took place. In addition, the adsorption bands at 696 cm⁻¹ and 847 cm⁻¹ (the libration modes of the two O–H⋯Cl hydrogen bonds)²⁷ were also lost, which were similar to the spectrum of β-FeOOH as shown in Fig. 4(b). The disappearance of the bond 847 cm⁻¹ was owing to the elimination of the chloride ions in SB@β-FeOOH precursor.

Fig. 4 The FTIR spectra of (a) SB, (b) β-FeOOH, (c) SB@β-FeOOH, and (d) SB@α-Fe₂O₃.

Table 1 The characteristic peaks of functional group assignments

Wave number/υ (cm^−1)	Functionality	Ref.
1640 cm^−1 in a	Stretching vibrations of the skeletal C=C	25
3200 to 3500 cm^−1 in b	O–H stretching vibrations	26 and 28
847 and 696 cm^−1 in b	Two O–H⋯Cl hydrogen bonds	27
696 cm^−1 in b	d-OH vibrations	27
1693 cm^−1 in b	H2O bending band	28
1429 and 1297 cm^−1 in b	Peculiar bonds to β-FeOOH	28
667, 588 and 441 cm^−1 in b	Characteristic Fe–O stretching modes	29
540 and 462 cm^−1 in d	Characteristic bands of α-Fe2O3	30

---

Based on the above discussion, we can summarize that the transformation from SB@β-FeOOH to SB@α-Fe₂O₃ through thermal decomposition process belongs to a structural change procedure. More specifically, water molecules, which weakly adsorbed onto the surface of β-FeOOH and can not be removed facially without changing their structure, were firstly eliminated during the thermal decomposition. As the reaction progresses, the dehydroxylation of the β-FeOOH took place and broke the parent chemical structure of β-FeOOH. Besides, the release of chloride is inevitable, hereby promoting the formation of α-Fe₂O₃ nanoparticles since the chloride ions are able to stabilize the chemical structure of β-FeOOH.³¹ Then, the re-crystallization from less crystalline to better crystallized α-Fe₂O₃ nanoparticles occurred. Finally, the stable residue can be transformed into composite SB@α-Fe₂O₃ catalysts.

DO adsorption enrichments over a fixed-bed column

With applications in human therapy and livestock industry, DO is one of the most widely used antibiotics in the world.³² Unfortunately, only 20–50% approximately of DO antibiotics can be absorbed into the organism. Vast majority of rudimental DO antibiotics get into surface water, groundwater, and other aquatic environment.³³ Exposures to low-level and accumulative DO antibiotics in the environment have drawn great attentions as it has a variety of potential dangerous effects, including impact on aquatic photosynthetic organisms, destruction of indigenous microbial populations, and dissemination into antibiotic-resistant genes among microorganisms.^34,35 Therefore, DO wastewater was targeted recently in the field of wastewater treatment.

The fixed-bed adsorption is easy to manipulate, low cost and can readily be scaled up. Generally, fixed-bed adsorption is implemented in a way that the influent wastewater comes in touch with a certain amount of sorbent, hereby affording room for management of large volume of flowing polluted fluid.³⁶ In view of those merits, the operation of adsorptive removal of DO antibiotics from aqueous solutions was carried out in a fixed-bed column using SB@α-Fe₂O₃ and bare SB as adsorbents.

The breakthrough curves of DO in the fixed-bed, filled with SB (Fig. 5(a)) and SB@α-Fe₂O₃ (Fig. 5(b)), respectively, were carried out on different EBCTs. In Fig. 5(a) and (b), a similar sharper breakthrough curve appeared at lower EBCTs for both SB and SB@α-Fe₂O₃ columns. It can be justified on account of the little contact time. In other words, at higher EBCTs, DO molecules had longer contact time in the adsorption column to diffuse into the pores of the SB@α-Fe₂O₃, resulting in a more effective adsorption removal of DO. In Fig. 5(c), the Q_e went up from 11.30 to 11.52 mg g⁻¹ for SB and from 11.38 to 12.86 mg g⁻¹ for SB@α-Fe₂O₃ with a growth of EBCT from 0.19 to 0.57 min. This phenomenon can be explained on the basis of residence time. That is, at lower EBCT, DO molecules would leave the adsorbents column before the adsorption equilibrium because of insufficient residence time, causing a reduction of adsorption capacity. The results were in agreement with those referred to the literature.³⁷

Fig. 5 Breakthrough curves representing DO adsorption onto (a) SB bed and (b) SB@α-Fe₂O₃ bed at EBCTs ranging from 0.19 to 0.57 min; and (c) equilibrium uptake capacity for both SB and SB@α-Fe₂O₃ at EBCTs ranging from 0.19 to 0.57 min (DO concentration: 10 mg L⁻¹).

The comparison of bare SB and SB@α-Fe₂O₃ fixed-bed columns for dynamic sorption of DO was implemented from the perspective of Q_e. As expected, in Fig. 5 (c), the Q_e value of SB@α-Fe₂O₃ was higher than that of unvarnished SB at all tested EBCTs. It indicated that SB@α-Fe₂O₃ columns had access to get better efficiency compared to the naked SB columns in terms of DO adsorption. The possible reasons are assigned to the embellishment of α-Fe₂O₃ nanoparticles on the biocarbon SB surface, which contributed to higher surface adsorption area and preferable adsorption performance. In other words, the enhanced removal of DO molecules was stemmed from the improved active adsorption sites on the surface of SB@α-Fe₂O₃. This consequence implied that it just needs fewer SB@α-Fe₂O₃ to deal with a given mass of DO than bare SB, making SB@α-Fe₂O₃ more cost effective.

The Thomas model is one of the most widely used models to present the performance theory of the adsorption process in fixed-bed systems. This model assumes plug flow behavior in the bed.³⁸ The model has the following linearized form:

where C₀ is the incipient concentrations (mg L⁻¹) of DO and C_t is the effluent concentrations (mg L⁻¹) of DO at any time t (min), k_T is the Thomas rate constant (mL min⁻¹ mg⁻¹), Q_e is the equilibrium uptake capacity (mg g⁻¹), M is the mass of adsorbent (g), and F is the flow rate (mL min⁻¹).

The eqn (4) was used to calculate the Thomas model parameters of the experimental data. The adsorption amount of the bed Q_e and the sorption kinetic rate constant k_T were calculated from a plot of ln[(C₀/C_t)] against t and the results were shown in Table 2. The experimental and predicted breakthrough curves of DO in the SB and SB@α-Fe₂O₃ fixed-bed columns were found to be consistent. The R² (regression coefficient) which ranged from 0.956 to 0.983 (SB) and from 0.945 to 0.986 (SB@α-Fe₂O₃) implies that the Thomas model described the experimental data for the adsorption of DO very well. Further, the values of k_T (mL min⁻¹ mg⁻¹) were 4.487, 7.679, and 10.808 for SB and 4.523, 7.384, 10.520 for SB@α-Fe₂O₃, respectively, with decreasing EBCT and adsorption capacity Q_e. The growth of k_T value means the reduction of the mass-transport resistance and axial dispersion. And, the decrease of Q_e value is evident since the adsorption capacity is in direct proportion to the contact time. Therefore, higher EBCT increase the rate of adsorption of DO in the SB@α-Fe₂O₃ column.

Table 2 Parameters of Thomas and Yoon–Nelson models for DO adsorption by SB@α-Fe₂O₃

	C0 (mg L^−1)	Z (cm)	EBCT (min)	Thomas model			Yoon–Nelson model
				kT × 10^−3 mL min^−1 mg^−1	Qe mg g^−1	R^2	kY min^−1	τ min	R^2
SB	10	2.0	0.57	4.487	11.100	0.983	0.045	111.169	0.983
	10	2.0	0.28	7.679	8.568	0.956	0.076	42.841	0.956
	10	2.0	0.19	10.808	7.354	0.983	0.108	24.512	0.983
SB@α-Fe2O3	10	2.0	0.57	4.523	13.188	0.980	0.045	131.878	0.980
	10	2.0	0.28	7.384	10.332	0.945	0.074	51.659	0.945
	10	2.0	0.19	10.520	8.826	0.986	0.105	29.419	0.986

---

The Yoon–Nelson model is a simple model and does not require detailed data about the type of adsorbent, the characteristics of adsorbate, and the physical properties of the adsorption bed.³⁹ This model is based on the assumption that the rate of decrease in the probability of adsorption for each adsorbate molecule is proportional to the probability of adsorbate adsorption and the probability of adsorbate breakthrough on the adsorbent.⁴⁰ The linearized Yoon–Nelson model regarding to a single component system can be expressed as eqn (5):

where k_Y is the rate velocity constant (min⁻¹), and τ is the time needed for 50% adsorbate breakthrough (min).

The kinetic coefficient k_Y and τ were determined by plotting ln[C_t/(C₀ − C_t)] versus t, with intercept τk_Y and slope of k_Y. The Yoon–Nelson model parameters were also shown in Table 2. As shown in Table 2, the Yoon–Nelson model adequately described the adsorption of DO on SB and SB@α-Fe₂O₃ in the fixed-bed columns. The R² values were all higher than 0.945. The times required for 50% DO breakthrough from the experiments matched the results of the τ received from the Yoon–Nelson model. The 50% breakthrough time τ decreased and rate constant k_Y increased when both EBCT decreased. In a comparison of R², we can see that both the Thomas and Yoon–Nelson models could be applied to predict the adsorption performance of DO in SB and SB@α-Fe₂O₃ fixed-bed column.

As a conventional rule, pH is a significant factor that influences the adsorption amounts of the adsorbent and has a profound effect on the nature of the physico–chemical interaction between the adsorbent and adsorbate in solutions. As can be seen from Fig. 6(a), the bare SB substrate exhibited zeta potential values 8.2, which is in line with those in literature that generally biocarbons are alkaline.^41,42 In other words, the content of acidic functional groups on the surface of SB biocarbon decreased as pyrolysis temperature went up,⁴³ whereas the content of total basic functional groups showed the opposite pattern. Therefore, at charring pyrolysis temperature of 773.15 K, the advanced loss of acidic surface functional groups and the increase of total surface basicity led to the alkaline surface of SB substrate. In contrast, the pH_PZC for SB@α-Fe₂O₃ samples decreased to 7.8 after the α-Fe₂O₃ nanoparticles were deposited onto. It can be noted hereby that the attachments of α-Fe₂O₃ nanoparticles on the surface of SB substrate has not resulted in a positive values of zeta potential, indicating the maintaining of basic functional groups of pristine SB. For DO molecules, DO is an amphoteric molecule as shown in Fig. 6(a), which exists as a DO⁺ cation below pH 3.5, as a DO⁰ zwitterion between pH 3.5 and 7.7, and as a DO⁻ or DO²⁻ anion, respectively, above pH 7.7 and 9.5. Consequently, below pH 3.5, the DO molecules have positive charge and so do SB@α-Fe₂O₃ samples. Above pH 8.0, DO molecules and SB@α-Fe₂O₃ were both negatively charged, which can not offer the strong affinity ability between DO molecules and SB@α-Fe₂O₃ adsorbent due to the presence of electrostatic repulsions, suggesting that the electrostatic attraction was not the overriding factor which affects the adsorption performance of SB@α-Fe₂O₃. Deducing from their own chemical structure, it seems that the strong π–π electron donor–acceptor interaction between the carbonaceous adsorbents and DO antibiotic drug molecules probably have played a dominant role during the adsorption procedure.^44,45 Therefore, in alkali solution, the weak adsorption capacity may root in the inferior interaction of π–π stacking or cation–π bonding with SB@α-Fe₂O₃ during the DO adsorption process.⁴⁶ Due to the buffering effect of the SB@α-Fe₂O₃, the weak DO adsorption amounts were also observed under near natural solution.⁴⁷

Fig. 6 The pH_PZC of the SB and SB@α-Fe₂O₃ samples (a) and breakthrough curves representing DO adsorption onto SB@α-Fe₂O₃ bed at different pH (b).

The significant effects of pH on the DO adsorption have also been ascertained by the variation of breakthrough curves of SB@α-Fe₂O₃ bed at different pH. In Fig. 6(b), the pH of DO solution fed to the fixed bed was on purpose set at 2.0, 4.0, 6.0, 8.0, and 11.0, respectively, while the initial DO concentration and EBCT were kept at 10 mg L⁻¹ and 0.57 min, respectively. The results displayed in Fig. 6(b) reveals that a raise in pH of the DO solution weakens the volume of contaminative water treated until breakthrough occurred. Furthermore, the steeper breakthrough curves were found with the increase in pH value. Meanwhile, the breakthrough curves transferred to a longer times, removing more DO molecules in the acidic condition.

Regeneration of DO-exhausted SB@α-Fe₂O₃ by H₂O₂

The subsequent regeneration and reuse of the adsorbent is crucial for the reduction of the adsorbent cost. Therefore, after the DO pre-adsorption, attempts were made to reuse and regenerate the DO-exhausted SB and SB@α-Fe₂O₃ by H₂O₂. Fig. 7 showed the breakthrough curves of DO adsorbed by SB and SB@α-Fe₂O₃ powder after in situ regenerated. Corresponding to EBCT (empty bed contact times) of 0.57, 0.28, 0.19 min, only 19.5, 21.2, 24.1% of the DO was absorbed by bare SB after H₂O₂-regeneration, respectively (the inset image in Fig. 7). It is generally known that activated carbon could catalyze the decomposing of H₂O₂ to create oxidant species. The pivotal mechanisms include that activated carbon possesses the ability to promote the degradation of some dissolved organic pollutants with the help of H₂O₂ through an electron-transfer.⁴⁸ Those indirect evidences suggest that SB bio-carbon would also facilitate the decomposition of H₂O₂ through an electron-transfer with SB and SB⁺ as the reduced and oxidized catalyst status,⁴⁹ respectively. But the decomposing procedures of H₂O₂ are used to decreasing with the appearance of adsorbed compounds.⁵⁰ Therefore, the adsorbed DO would block the catalytic active sites of the SB surface, refraining the catalytic performance of the bare SB surface toward the H₂O₂ decomposition. Thus, the adsorption capacity of the original SB was far from being restored and recycled. However, compared with the undecorated SB, spent SB@α-Fe₂O₃ has achieved the higher regeneration efficiency η with the help of H₂O₂. The precise regeneration efficiency value of DO-saturated SB@α-Fe₂O₃ tests was 90.06, 89.99, 83.28%, which can be assigned to the coupling effect. The coupling effect of SB bio-carbon catalytic oxidation and α-Fe₂O₃-triggered heterogeneous Fenton-like oxidation was in action synchronously. Hereby the composite adsorbent SB@α-Fe₂O₃ could be effectively restored. Moreover, it was confirmed that regeneration efficiency η of SB@α-Fe₂O₃ had improvements of 70.56, 68.79 and 59.18% than that of bare SB, demonstrating that the α-Fe₂O₃-triggered heterogeneous Fenton-like oxidation guide the dominant contributor in the regenerative system. More detailedly, owing to the high catalytic activity of α-Fe₂O₃, the constructed α-Fe₂O₃ heterogeneous Fenton-like system could produce much more free radicals, such as ˙OH, ˙OOH and ¹O₂. These active free radicals possess outstanding oxidation ability to degrade the DO compounds, hereby achieving the remarkable regeneration of SB@α-Fe₂O₃ sorbent. A schematic diagram is shown in Scheme 1.

Fig. 7 Breakthrough curves of in situ regenerated SB and SB@α-Fe₂O₃ at EBCTs ranging from 0.19 to 0.57 min (DO concentration: 10 mg L⁻¹; H₂O₂ concentration 1% w/v).

Scheme 1 Schematic of the regeneration mechanism of SB@α-Fe₂O₃ in the removal of DO.

In order to better exemplify the above-mentioned α-Fe₂O₃-triggered heterogeneous Fenton-like regeneration mechanism, t-butanol,⁵¹ p-benzoquinone⁵² and sodium azide⁵³ were introduced on purpose in this work as scavengers of ˙OH, ˙OOH and ¹O₂, separately. The experimental results showed that regeneration efficiency η was decreased to 42.18, 49.06 and 33.91%, respectively, in the presence of t-butanol, p-benzoquinone and sodium azide. Compared with the parallel degradation process of DO, the regeneration efficiency η were reduced by 47.88, 41.00 and 56.15%. The decrease of regeneration efficiency η in the presence of t-butanol suggested that the hydroxyl radical ˙OH might be one of the radical intermediates. In like manner, ˙OOH radical was also proved to be an active radical species since the p-benzoquinone (the scavenger of ˙OOH) deminished the regeneration efficiency η. In contrast, the regeneration was not completely hampered by the addition of t-butanol or p-benzoquinone. This means that there are other active reaction intermediates drawn into the procedure which could not be quenched by t-butanol or p-benzoquinone. Interestingly, in the case of sodium azide, the η of SB@α-Fe₂O₃ went down even more than first two cases, which declared that the ¹O₂ plays a pivotal role in the regeneration of SB@α-Fe₂O₃. With the aid of the reported research literature, the detailed description of regeneration procedure in the presence of t-butanol, p-benzoquinone and sodium azide can be induced as follows:

According to reported literature, hydroxyl radical ˙OH has an average lifetime of 200 ns,⁵⁴ which is consistent with a diffusion distance of about 20 nm. In the present work, for α-Fe₂O₃ with particle size much larger than 20 nm, ˙OH generated on the surface of SB@α-Fe₂O₃ would partially encounter a limit to reach the regeneration sites. Thus, ¹O₂ produced from ˙OH should be more highly valued than ˙OH. In addition, the disclosed inhibition effect of t-butanol and p-benzoquinone could be explained by the fact that totally capture of the ˙OH or ˙OOH would only partially break off the formation of ¹O₂. The relevant radical reactions were estimated as shown in eqn (6)–(13).^55,56 Based on these reactions, ˙OH, ˙OOH and ¹O₂ were all participated in the degradation of DO pollutants. In these ways, the DO-saturated adsorption sites on the surface of SB@α-Fe₂O₃ composites could be easily regenerated.

SB≡Fe³⁺ + H₂O₂ → SB≡Fe²⁺ + ˙OOH + H⁺ (6)

SB≡Fe²⁺ + H₂O₂ →SB≡Fe³⁺ + ˙OH + OH⁻ (7)

˙OH + H₂O₂ → ˙OOH + H₂O (8)

˙OOH → H⁺ + O₂˙⁻ (9)

O₂˙⁻ + ˙OH → ¹O₂ + OH⁻ (10)

˙OOH + O₂˙⁻ → ¹O₂ + HOO⁻ (11)

˙OOH + ˙OOH → ¹O₂ + H₂O₂ (12)

Radicals + DO → degradation products (13)

In order to further evaluate the durative reuse of SB@α-Fe₂O₃ for practical application, the durability of the SB@α-Fe₂O₃ composites were assessed by employing the DO-saturated SB@α-Fe₂O₃ which were treated by H₂O₂ (1% w/v). For three consecutive cycles, the η of SB@α-Fe₂O₃ decreased from 90.06% to 79.84%, which also coincided with an alike decrease of the adsorption ability during the first and second adsorption cycles. The conceivable reason for this might be that the repulsion by the retained DO from previous cycles or stronger affinity of the DO intermediate products produced during the degradation process. Overall, the results indicated that SB@α-Fe₂O₃ integrated the catalytic properties of α-Fe₂O₃ nanoparticles, implying that the SB@α-Fe₂O₃ could be used on multiple adsorption cycles.

Moreover, the longevity of the SB@α-Fe₂O₃ composites were predicted by using three parameters, including the uptake capacity (Q), breakthrough time (t_b) and critical bed height (Z) respectively. Results were illustrated in Fig. 8 and linear regression equations were determined as:

t_b = t_i + k_bN (14)

Z = Z_i + k_ZN (15)

Q = Q_i + k_QN (16)

where t_i, Z_i and Q_i are the initial breakthrough time (min), critical bed height (cm) and initial column uptake (mg g⁻¹), respectively; k_b, k_Z and k_Q are the matching life factor of the breakthrough time, critical bed height and column uptake, separately; besides, N delegates the cycle number.

Fig. 8 Linear plots of column uptake, critical bed height and breakthrough time with regard to number of cycles.

As shown in Fig. 8, the values of t_i and k_b were acquired to be 55.0 min and 10.0 min per cycle, respectively, predicting 5.5 cycles of DO-saturated SB@α-Fe₂O₃ bed which had sufficient capacity to avert the breakthrough at time t = 0 for up. Moreover, the linear regression equation Q = 14.17 − 1.3N was also formulated and it was effortless to estimate that the column bed would be exhausted after 10.9 cycles. Additionally, the value of Z_i was concluded to 2.17 cm and k_Z = 0.15 cm per cycle. It means that the breakthrough would emerge at t = 0 after 14.5 cycles. Thus, it can be determined that the SB@α-Fe₂O₃ column bed would be fully exhausted after 5 cycles, and the sorption zone will not arrive at the top when the SB@α-Fe₂O₃ column is totally exhausted. From these results, the adsorbent of SB@α-Fe₂O₃ can be commendably regenerated by heterogeneous Fenton-like system, and the SB@α-Fe₂O₃ could be reused several times for DO wastewater purification with a low running expenditure, confirming the excellent regeneration ability and long-term stability of SB@α-Fe₂O₃ composites.

The separable performances of resultant catalysts were firstly examined from the macroscopic point of view using magnet. As seen in Fig. 9(a), the SB@α-Fe₂O₃ samples were dispersed previously in the deionized water, forming a stable black suspension. When an external magnet was closed to the cuvette, the suspended samples were not attracted (Fig. 9(b)), suggesting that nonmagnetic properties were endowed with the SB@α-Fe₂O₃ species. However, after several minutes, it is obvious that the natural sedimentation occurred within the cuvette due to gravity (Fig. 9(c)), indicating that the SB@α-Fe₂O₃ composites can be easily separated from the solution via common filtering or centrifugation methods.

Fig. 9 Photographs of the dispersion and separation of SB@α-Fe₂O₃: (a) without external magnetic field, (b) with external magnetic field and (c) natural sedimentation.

Conclusions

In summary, a novel and low-cost SB@α-Fe₂O₃ composites were successfully synthesized through a simple thermal conversion process in this work. The performance of SB@α-Fe₂O₃ and SB fixed-bed columns was investigated on the adsorption of DO from aqueous solution. The SB@α-Fe₂O₃-bed column demonstrated greater adsorption ability than the SB-bed column treating 10 mg L⁻¹ DO solution. Heterogeneous Fenton-like oxidation proved to be a simple and efficient process for in situ regeneration of the DO-exhausted SB@α-Fe₂O₃. The effective way of regeneration and the robustness of the adsorbent SB@α-Fe₂O₃ make it possible for consecutive adsorption/regeneration cycles with a low running cost. Considering the facile synthetic method of composites, the superior adsorption performance and the effective way of regeneration for consecutive pre-adsorption/regeneration cycles, we can speculate that the SB@α-Fe₂O₃ would be a promising and practical adsorbent for the removal and destruction of pharmaceuticals-contaminated waters. The present strategy of embedding α-Fe₂O₃ nanoparticles onto the waste SB can be also extended to the simple fabrication of other α-Fe₂O₃ nanoparticles/biocarbon materials with similar structure.

Acknowledgements

This work was supported by China Postdoctoral Science Special Foundation, Scientific Research Foundation for the Returned Overseas Chinese Scholars, National Natural Science Foundation of China (No. 21176031), Shanxi Provincial Natural Science Foundation of China (No. 2015JM2071) and the Fundamental Research Funds for the Central Universities (No. 310829162014).

References

1. L. N. Shao, Z. M. Ren, G. S. Zhang and L. L. Chen, Mater. Chem. Phys., 2012, 135, 16–24.
2. W. G. Li, X. J. Gong, K. Wang, X. R. Zhang and W. B. Fan, Bioresour. Technol., 2014, 165, 166–173.
3. J. Herney-Ramirez, M. A. Vicente and L. M. Madeira, Appl. Catal., B, 2010, 98, 10–26.
4. Y. J. Yao, J. C. Qin, Y. M. Cai, F. Y. Wei, F. Lu and S. B. Wang, Environ. Sci. Pollut. Res., 2014, 21, 7296–7306.
5. M. B. Kasiri, H. Aleboyeh and A. Aleboyeh, Appl. Catal., B, 2008, 84, 9–15.
6. L. L. Ding, B. Zou, W. Gao, Q. Liu, Z. C. Wang, Y. P. Guo, X. F. Wang and Y. H. Liu, Colloids Surf., A, 2014, 446, 1–7.
7. M. C. Ribas, M. A. Adebayo, L. D. T. Prola, E. C. Lima, R. Cataluña, L. A. Feris, M. J. Puchana-Rosero, F. M. Machado, F. A. Pavan and T. Calvete, Chem. Eng. J., 2014, 248, 315–326.
8. H. Deng, L. Yang, G. H. Tao and J. L. Dai, J. Hazard. Mater., 2009, 166, 1514–1521.
9. M. Ahmad, S. S. Lee, X. M. Dou, D. Mohan, J. K. Sung, J. E. Yang and Y. S. Ok, Bioresour. Technol., 2012, 118, 536–544.
10. L. D. Kagliwal, S. C. Patil, A. S. Pol, R. S. Singhal and V. B. Patravale, Sep. Purif. Technol., 2011, 80, 533–540.
11. S. Ibrahim, W. Z. Shuy, H. M. Ang and S. Wang, Asia-Pac. J. Chem. Eng., 2010, 5, 563–569.
12. R. Song, B. Bai, G. L. Puma, H. L. Wang and Y. R. Suo, React. Kinet. Mech. Cat., 2015, 115, 547–562.
13. X. Zhang, B. Bai, G. L. Puma, H. L. Wang and Y. R. Suo, Chem. Eng. J., 2016, 284, 698–707.
14. Y. W. Gao, Y. Z. Guo and H. Zhang, J. Hazard. Mater., 2016, 302, 105–113.
15. D. A. Wheeler, G. Wang, Y. Ling, Y. Li and J. Z. Zhang, Energy Environ. Sci., 2012, 5, 6682–6702.
16. Z. N. Liu, X. X. Zhang and L. Zhao, Adv. Mater. Res., 2012, 518, 2414–2417.
17. C. M. Zheng, X. Z. Chen, C. W. Yang, C. J. Zhang, H. L. Li, L. X. Kan, J. Xia and X. H. Sun, RSC. Adv., 2015, 5, 98842–98852.
18. I. Ursachi, A. Stancu and A. Vasile, J. Colloid Interface Sci., 2012, 377, 184–190.
19. W. J. Liu, F. X. Zeng, H. Jiang and X. S. Zhang, Bioresour. Technol., 2011, 102, 8247–8252.
20. K. M. Parida, J. Mater. Sci., 1988, 23, 1201–1205.
21. D. Stoyanova, R. Nickolov, M. Khristova, D. Paneva and D. Mehandjiev, J. Porous Mater., 2009, 16, 1–7.
22. Z. C. Kadirova, K. I. Katsumata, T. Isobe, N. Matsushita, A. Nakajima and K. Okada, J. Environ. Chem. Eng., 2014, 2, 2026–2036.
23. H. C. Lan, A. M. Wang, R. P. Liu, H. J. Liu and J. H. Qu, J. Hazard. Mater., 2015, 285, 167–172.
24. C. S. Castro, M. C. Guerreiro, L. C. A. Oliveira, M. Gonçalves, A. S. Anastácio and M. Nazzarro, Appl. Catal., A, 2009, 367, 53–58.
25. T. Wan, R. Q. Huang, Q. H. Zhao, L. Xiong, L. L. Qin, X. M. Tan and G. J. Cai, J. Appl. Polym. Sci., 2013, 130, 3404–3410.
26. O. Glemser and G. Rieck, Z. Anorg. Allg. Chem., 1958, 297, 175–188.
27. E. A. Deliyanni, D. N. Bakoyannakis, A. I. Zouboulis, K. A. Matis and L. Nalbandian, Microporous Mesoporous Mater., 2001, 42, 49–57.
28. K. Kauffman and F. Hazel, J. Inorg. Nucl. Chem., 1975, 37, 1139–1148.
29. T. Sugimoto and A. Muramatsu, J. Colloid Interface Sci., 1996, 184, 626–638.
30. Y. S. Wang, A. Muramatsu and T. Sugimoto, Colloids Surf., A, 1998, 134, 281–297.
31. J. Ellis, R. Giovanoli and W. Stumm, Chimia, 1976, 30, 194–197.
32. M. Majewski, Pol. Ann. Med., 2014, 21, 57–62.
33. X. Zhao, Z. L. Chen, X. C. Wang, J. C. Z. Li, J. M. Shen and H. Xu, Bioresour. Technol., 2015, 179, 104–112.
34. Y. P. Zhao, F. Tong, X. Y. Gu, C. Gu, X. R. Wang and Y. Zhang, Sci. Total Environ., 2014, 470, 19–25.
35. X. L. Bao, Z. M. Qiang, J. H. Chang, W. W. Ben and J. H. Qu, J. Environ. Sci., 2014, 26, 962–969.
36. S. K. Maji, Y. H. Kao, C. J. Wang, G. S. Lu, J. J. Wu and C. W. Liu, Chem. Eng. J., 2012, 203, 285–293.
37. V. C. Taty-Costodes, H. Fauduet, C. Porte and Y. S. Ho, J. Hazard. Mater., 2005, 123, 135–144.
38. H. C. Thomas, J. Am. Chem. Soc., 1944, 66, 1466–1664.
39. M. Calero, F. Hernainz, G. Blazquez, G. Tenorio and M. A. Martin-Lara, J. Hazard. Mater., 2009, 171, 886–893.
40. S. S. Baral, N. Das, T. S. Ramulu, S. K. Sahoo, S. N. Das and G. Roy Chaudhury, J. Hazard. Mater., 2009, 161, 1427–1435.
41. M. Jouiad, N. Al-Nofeli, N. Khalifa, F. Benyettou and L. F. Yousef, J. Anal. Appl. Pyrolysis, 2015, 111, 183–190.
42. X. P. Gai, H. Y. Wang, J. Liu, L. M. Zhai and S. Liu, PLoS One, 2014, 9, e113888.
43. M. I. Al-Wabel, A. Al-Omran, A. H. El-Naggar, M. Nadeem and A. R. A. Usman, Bioresour. Technol., 2013, 131, 374–379.
44. W. Yang, F. Zheng, Y. Lu, X. Xue and N. Li, Ind. Eng. Chem. Res., 2011, 50, 13892–13898.
45. L. Ji, W. Chen, L. Duan and D. Zhu, Environ. Sci. Technol., 2009, 43, 2322–2327.
46. Y. Gao, Y. Li, L. Zhang, H. Huang, J. Hu, S. M. Shah and X. Su, J. Colloid Interface Sci., 2012, 368, 540–546.
47. P. Liu, W. J. Liu, H. Jiang, J. J. Chen, W. W. Li and H. Q. Yu, Bioresour. Technol., 2012, 121, 235–240.
48. A. Rey, J. A. Zazo, J. A. Casas, A. Bahamonde and J. J. Rodriguez, Appl. Catal., B, 2011, 402, 146–155.
49. A. Georgi and F. D. Kopinke, Appl. Catal., B, 2005, 58, 9–18.
50. A. Bach and R. Semiat, Desalination, 2011, 273, 57–63.
51. J. Benner and T. A. Ternes, Environ. Sci. Technol., 2009, 43, 5472–5480.
52. M. C. Yin, Z. S. Li, J. H. Kou and Z. G. Zou, Environ. Sci. Technol., 2009, 43, 8361–8366.
53. L. K. Ge, J. W. Chen, X. L. Qiao, J. Lin and X. Y. Cai, Environ. Sci. Technol., 2009, 43, 3101–3107.
54. K. Rusevova, F. D. Kopinke and A. Georgi, J. Hazard. Mater., 2012, 241, 433–440.
55. W. G. Li, Y. Wang and A. Irini, Chem. Eng. J., 2014, 244, 1–8.
56. L. Zhou, W. Song, Z. Q. Chen and G. C. Yin, Environ. Sci. Technol., 2013, 47, 3833–3839.

---