Promoting desulfurization capacity and separation efficiency simultaneously by the novel magnetic Fe₃O₄@PAA@MOF-199

Abstract

The world is becoming more stringent on lowering the sulfur concentrations in fuels. To fulfill this expectation, a new type of magnetic desulfurization adsorbent, Fe₃O₄@PAA@MOF-199, was designed and fabricated using a facile two-step assembly approach, in which PAA inventively acted like a bridge to incorporate different amounts of magnetite Fe₃O₄ into a MOF-199 crystal matrix. Fe₃O₄@PAA@MOF-199s were demonstrated to be efficient adsorbents for the removal of S-compounds, such as thiophene, benzothiophene (BT) and dibenzothiophene (DBT), from a model fuel, and the sulfur saturated adsorption capacity followed the order of DBT > BT > thiophene. The magnetization of Fe₃O₄@PAA@MOF-199s ensured that the adsorbents had good performance in magnetic separation. The relative high adsorption capacity, the separation efficiency, as well as the stable recyclability indicated that magnetic Fe₃O₄@PAA@MOF-199 would be a promising adsorbent in adsorptive desulfurization.

---

Introduction

The environmental legislation on SOx exhaust levels has become increasingly stringent worldwide. It is very important to lower the sulfur concentrations in fuels to 10 ppmw S (parts per million by weight of sulfur) or less.^1–4 Currently, the main industrial process is hydrodesulfurization (HDS), in which sulfur compounds such as thiophene, benzothiophene (BT), and dibenzothiophene (DBT) in fuel feeds, are hydrogenated to hydrocarbons and H₂S. However, this conventional process is not suitable for ultra-deep desulfurization because of highly demanded operation conditions and cost, such as high temperatures and pressures, extra consumption of H₂, and serious deactivation of the catalysts. In addition, refractory aromatic sulfur compounds, especially those with the steric hindrance, like DBT and 4,6-dimethyldibenzothiophene (4,6-DMDBT), are inefficient to remove by HDS because of their low reactivity in a low concentration.⁵

Compared with HDS technology, a promising way to remove S-compounds would be adsorption because it can be carried out under ambient temperature and pressure. More importantly, it has the capability to reduce the sulfur content to less than 1 ppmw S.⁶ Various adsorbents, including zeolites,^7–9 and activated carbons,^10–12 have been explored over the years. Generally, efficient desulfurization by adsorption comes from the strong interaction between S-compounds and the adsorbent. It has been proposed that incorporating transition metal ions, such as Cu⁺ and Ag⁺, into the microporous materials could result in high S-adsorption capacity and high selective desulfurization, which were attributed to the formation of π-complexation between S-compounds and metal ions, as well as direct sulfur−metal (S−M) interaction.^13,14

Metal-organic frameworks (MOFs) are a promising type of adsorbent because of their highly ordered three-dimensional porous networks, high inner surface areas and large pore volumes. The desulfurization capacity of MOFs was found to be determined by the pore size and shape.^15,16 Li et al.¹⁷ investigated four types of MOFs, and they found that the adsorption capacity for DBT follows the order of Cu–BTC > Cr–BDC > Cr–BTC ≫ Cu–BDC, which was considered to be a result of the comprehensive effects, such as the suitable pore size and shape, the framework structure, as well as the exposed Lewis acid sites at the surface of the pore. Jhung et al.¹⁸ embedded CuCl₂ into porous MIL-47 (vanadium-benzenedicarboxylate) and achieved a remarkably high saturation adsorption capacity (310 mg BT g⁻¹ at 25 °C). Among the numerous MOFs reported to date, one of the most popular adsorbents is the porous copper–benzenetricarboxylate (Cu–BTC, i.e. MOF-199, HKUST-1 or C300), which has been realized in mass production.¹⁹ Herein, we proposed the use of MOF-199 for the selective adsorption of S-compounds.

For the technique of the desulfurization process, the magnetic separation of adsorbents from fuel based on the superparamagnetic particles is obviously much more convenient and efficient. In fact, magnetic separation has been used in diverse areas,^20–25 which provide a good base for us to design magnetic MOF nanocomposites. Very recently, magnetic MOF/Fe₃O₄ nanohybrids have been successfully fabricated through a layer by layer (LBL) method.²⁶ However, the complicated operation procedure, high solvent consumption, and low coating thickness limit its use in real applications. Kaskel et al. prepared superparamagnetic functionalized MOFs by integrating superparamagnetic iron-oxide nanoparticles into polycrystalline MOF aggregates.²⁷ Although the concept allows the external manipulation of highly microporous MOFs for efficient catalyst separation, the obtained composite material was heterogeneous and hence showed lower separation efficiency.

In the present study, a novel facile synthesis method of magnetic Fe₃O₄@PAA@MOF-199 was proposed. Polyacrylic acid (PAA) chains in this composite acted like a bridge, connecting the inside magnetite nanoparticles Fe₃O₄ and outside MOF-199 layer. On the one hand, the carboxyl group in the PAA chain can strongly coordinate with ferric ions as a binding to produce highly uniform magnetite submicrospheres; however, it can simultaneously coordinate with copper ions as a substrate for the further growth of MOF-199. The desulfurization performances of the magnetic adsorbent were tested for removing thiophene, BT and DBT in n-octane solvent. The efficiency of magnetic separation and the recycling desulfurization performance were also investigated.

Experimental section

Chemicals

Analytical grade reagents, such as ferric chloride (FeCl₃·6H₂O, ≥99.0%), sodium acetate (NaAc, ≥99.0%), ethylene glycol (EG, ≥99.0%), copper nitrate (Cu(NO₃)₂·3H₂O, ≥99.0%), n-Octane (≥95.0%), were purchased from Sinopharm Chemical Reagent Co., Ltd. Polyacrylic acid (PAA, M_w ≈ 3000), p-xylene (≥99.0%), and thiophene (99%) were purchased from Aladdin Chemistry Co. 1,3,5-Benzenetricarboxylic acid (H₃BTC, ≥95%), benzothiophene (≥98%) and dibenzothiophene (≥98%) were purchased from Sigma-Aldrich Co. All chemicals were used as purchased without further purification.

Synthesis

A two-step assembly approach was adopted for the synthesis of Fe₃O₄@PAA@MOF-199. As shown in Scheme 1, the first step was the assembly of FeCl₃ and the PAA chain, and then the second step was the assembly of Fe₃O₄@PAA submicrospheres and MOF precursors.

Scheme 1 Possible formation approach of the magnetic Fe₃O₄@PAA@MOF-199 microspheres.

During the first assembly processes, Fe³⁺ ions were first coordinated with carboxyl groups on PAA chains, and then hydrolyzed to Fe(OH)₃ by alkaline NaAc, and also partially reduced to Fe(OH)₂ by ethylene glycol. Fe(OH)₃ and Fe(OH)₂ were further dehydrated and formed magnetite Fe₃O₄ when heating at 200 °C. Based on this, Fe₃O₄@PAA was synthesized using a modified procedure reported by Liang et al.²⁸ Typically, 1.47 g FeCl₃·6H₂O was first completely dissolved in 80 mL ethylene glycol. When 1.04 g PAA was added, a golden yellow aqueous solution was obtained. After stirring for 30 min, 5.40 g NaAc was added to the solution. After ultrasonic mixing for 30 min, the whole reaction solution was placed into a 100 mL Teflon-lined stainless-steel autoclave and heated at 200 °C for 24 h. The synthesized magnetite submicrospheres were washed with ethanol and deionized water three times and separated using a magnet. The final product was dispersed and stored in 80 mL ethanol.

During the second assembly processes, Cu²⁺ ions were coordinated with carboxyl groups of PAA chains on the surface of Fe₃O₄@PAA submicrospheres, and then the unsaturated Cu²⁺ ions were further coordinated with carboxyl groups of H₃BTC for the growth of the first seed layer of MOF-199 crystals. Then, the MOF-199 crystals would grow up around Fe₃O₄@PAA submicrospheres. Moreover, instead of growing around the Fe₃O₄@PAA submicrospheres, MOF-199 crystals may also form in the bulk solution. For the synthesis details, 1 g Cu(NO₃)₂·3H₂O and 0.5 g H₃BTC were completely dissolved in 60 mL EtOH. Then, a 5 mL or 15 mL as-synthesized Fe₃O₄@PAA ethanol solution was added to the MOF-199 precursor solution, respectively. After stirring at 80 °C for 2 h, the products of Fe₃O₄@PAA@MOF-199 (L, low Fe₃O₄ content) or Fe₃O₄@PAA@MOF-199 (H, high Fe₃O₄ content) were finally obtained after washing with EtOH, separating through a magnet, and drying at 100 °C. Pure MOF-199 without adding Fe₃O₄@PAA was also synthesized in similar manner for comparison.

Characterization

The morphology of samples was characterized by using a Nova NanoS 450 field emission scanning electron microscope (FESEM) operated at 5 kV. All transmission electron microscopy images were obtained by using a JEOL JEM-2100 instrument. The powder X-ray diffraction (PXRD) data were collected on a D/Max2550 VB/PC diffractometer (40 kV, 200 mA) using a Cu Kα as the radiation. N₂ adsorption–desorption isotherms at 77 K were measured by a volumetric adsorption analyzer ASAP 2020. The specific surface area was calculated by the Brunauer–Emmett–Teller (BET) method. The FTIR spectra were carried out on a Nicolet iS10 FTIR spectrometer using the KBr pellet technique. The thermal stability was detected using a TGA unit (NETZSCH STA 499 F3). About 10 mg of the sample was heated from 25 °C to 600 °C at a heating rate of 10 °C min⁻¹ in nitrogen with a flow rate of 40 mL min⁻¹. X-ray photoelectron spectroscopy (XPS) measurement was conducted with a PHI 5000C ESCA spectrometer (Perkin-Elmer, USA) with an Al Kα radiation as the X-ray source. Magnetization of the samples were carried out on a Lakeshore 7407 vibration sample magnetometer (VSM). Inductively coupled plasma atomic emission spectrometry (ICP-AES, Vanan 710) was employed to determine the copper(II) content in the samples, and the samples were treated with a mixture of HCl and HNO₃ before the measurements.

Desulfurization measurements

The desulfurization performance of Fe₃O₄@PAA@MOF-199s was tested by the adsorption capacity of sulfur in the model oil, which was prepared by dissolving thiophene, BT or DBT in n-octane. With a ratio of (0.02 g adsorbent)/(5 mL oil), all adsorption experiments were conducted in a 10 mL glass bottle at 25 °C with stirring. Prior to adsorption, each adsorbent was heated in a vacuum at 150 °C for 2 h. The concentration of sulfur in oil was determined by a gas chromatography-flame photometric detector (GC-FPD, GC-950, Haixin Chromatography), which was equipped with a HP-5 capillary column (15 m × 0.53 mm × 1.5 μm film thickness). The adsorption capacity was calculated bywhere Q_i is the adsorption capacity of sulfur adsorbed on the adsorbent (mg S g⁻¹ adsorbent), W is the mass of model oil (g), M is the mass of the adsorbent used (g), and C₀ and C_i are the initial and final concentrations of sulfur in the model oil (mg g⁻¹), respectively.

The saturated adsorption capacity was calculated using the Langmuir adsorption model when the adsorption equilibrium was reached. The adsorption isotherms of Fe₃O₄@PAA@MOF-199s can be plotted according to the Langmuir equation (eqn (2)),

where C_e is the equilibrium concentration of S-compound (mg L⁻¹), Q_e is the amount of sulfur adsorbed at equilibrium (mg S g⁻¹), Q₀ is the saturated adsorption capacity (mg S g⁻¹), b is the Langmuir constant (L mg⁻¹). Therefore, the maximum adsorption capacity Q₀, can be obtained from the reciprocal of the slope of the plot of C_e/Q_e against C_e.

Recyclability and regeneration

The separation performance of the adsorbent was determined on a scattering turbidimeter (WGZ-800, Shanke instrument) by measuring the turbidity of the oil after the adsorbent was separated by the magnet. The turbidity of the oil with pure MOF-199 was also determined in a similar way for comparison.

Regeneration of the adsorbent was performed by solvent extraction. After adsorption and magnetic separation, the adsorbent was extracted by p-xylene three times to remove the adsorbed S-compounds, and then was dried at 150 °C overnight. The recycling desulfurization performance was tested by taking the regenerated adsorbent as a fresh adsorbent. The recycling desulfurization was repeated 5 times.

Results and discussion

Characterization of magnetic MOFs

The magnetic composites of Fe₃O₄@PAA@MOF-199s were obtained by a two-step self-assembly approach. The crystalline structures and phase composition of Fe₃O₄@PAA submicrospheres and Fe₃O₄@PAA@MOF-199 were characterized by PXRD. As shown in Fig. 1, six diffraction peaks in the pattern of Fe₃O₄@PAA at 2θ = 30.2° (220), 35.7° (311), 43.4° (400), 53.7° (422), 57.4° (511), and 62.9° (440) can be indexed to a typical phase of Fe₃O₄, which are in good agreement with the characteristic peaks of a standard magnetite crystal (JCPDS no. 19-0629, isometric-hexoctahedral crystal system). In addition, XPS was performed to analyze the chemical states of iron (ESI, Fig. S1†). The peaks at about 710 eV and 720 eV, corresponding to Fe 2p3/2 and Fe 2p1/2, confirms that the obtained submicrospheres consist of pristine Fe₃O₄. The FTIR spectrum of Fe₃O₄@PAA (ESI, Fig. S2†) also demonstrated the formation of the composite, in which the absorption band at about 1452 cm⁻¹ correspond to the –CH₂– bending vibration of the PAA chain, the absorption at about 1715 cm⁻¹ is indicative of C=O stretching, and the absorption bands at about 1571 cm⁻¹ and 1407 cm⁻¹ are assigned to the asymmetric and symmetric stretching bands of COO⁻, respectively. Based on the calculations using the Debye–Scherrer formula for the strongest (311) diffraction peak, the size of Fe₃O₄ nanocrystals was ca. 15 nm. Compared with the PXRD pattern of pristine MOF-199, Fe₃O₄@PAA@MOF-199 (H) shows almost the same diffraction patterns, indicating MOF-199 structure was well developed. Because of the limited amount of Fe₃O₄ in Fe₃O₄@PAA@MOF-199 (H), as well as overlap of the diffraction peaks of Fe₃O₄ and MOF-199, the diffraction peaks of Fe₃O₄ were not clearly observed. Moreover, the FTIR spectra of Fe₃O₄@PAA, MOF-199 and Fe₃O₄@PAA@MOF-199 (ESI, Fig. S2†) further demonstrate the successful formation of Fe₃O₄@PAA@MOF-199 composites.

Fig. 1 X-ray diffraction patterns of Fe₃O₄@PAA, MOF-199 and Fe₃O₄@PAA@MOF-199 (H).

Fig. 2 shows the morphology of the magnetic particles. The FESEM images of Fe₃O₄@PAA (Fig. 2a) reveal that the submicrospheres obtained by the solvothermal method are uniform, with a mean size of about 600 nm. Each Fe₃O₄@PAA microsphere consists of small Fe₃O₄ nanocrystals, which are enwrapped by PAA chains, and finally aggregated to large submicrospheres due to the tendency of reducing the strong surface tension. The SEM image of Fe₃O₄@PAA@MOF-199 (Fig. 2b) shows that Fe₃O₄@PAA submicrospheres are embedded in the continuous framework of MOF-199 crystals. In contrast, its TEM image (Fig. 2c and d) further reveals that MOF-199 crystals grow up at the surface of Fe₃O₄@PAA submicrospheres and these submicrospheres are distributed in the MOF-199 matrix. Fig. 3 shows the bright-field TEM image of Fe₃O₄@PAA@MOF-199 and its corresponding EDS elemental mapping of Cu, C, Fe, and O. With denser Cu and C closely surrounding the Fe₃O₄ core, it confirms that Fe₃O₄@PAA submicrospheres are indeed embedded in MOF-199 crystals. On the other hand, the O mapping image indicates that it is distributed over the materials that are relatively denser in the core due to the contributions of both Fe₃O₄ and PAA. More importantly, regardless of the elemental mapping of C or of O, both distributions are even without voids between the Fe₃O₄@PAA core and MOF-199 shell, indicating that MOF-199 crystals are closely packed around the surface of the Fe₃O₄@PAA submicrospheres, and further confirms the formation mechanism, as proposed in Scheme 1, in which carboxyl groups in the PAA chain can coordinate with Cu²⁺ cations to form the first seed layer of MOF-199, and induce the further growth of MOF-199 crystals around the core of Fe₃O₄@PAA. To illustrate the important role of PAA in the formation of the composite, we dispersed Fe₃O₄ nanoparticles and Fe₃O₄@PAA submicrospheres separately into a Cu(NO₃)₂ solution. The ICP-AES results, as listed in Table S1,† show that there are almost no copper ions absorbed in Fe₃O₄, whereas the content of copper ions in Fe₃O₄@PAA was as high as 80 mg g⁻¹, suggesting that the carboxyl group in PAA chain can strongly coordinate with copper ions. The closely attached copper ions could form the first seed layer of MOF-199 at the surface of the Fe₃O₄@PAA submicrospheres. Hence, the further growth of MOF-199 crystals can be achieved. This following growth of MOF-199 was similar to the Fe₃O₄@PAA@MOF-199 submicrospheres (ESI, Fig. S3†) obtained by the complex LBL method, which can further illustrate the reality of our formation mechanism. However, in some of these works, Fe₃O₄@PAA submicrospheres were not totally coated with MOF-199 crystals due to the orientated growth of MOF crystals in one-pot synthesis. Nevertheless, we achieved significantly improved synthesis conditions, including solvent changes from the expensive and hazardous DMF to a common solvent such as ethanol, and the reaction time was greatly shortened to 2 h compared with the normal reaction time of 20 h (ref. 30) to meet the requirements of future industrial applications. In fact, just because of these growth defects, some parts of the magnetite Fe₃O₄@PAA were exposed, which would present stronger magnetism for magnetic separation.

Fig. 2 SEM images of Fe₃O₄@PAA (a), Fe₃O₄@PAA@MOF-199 (b) and TEM images of Fe₃O₄@PAA@MOF-199 (c and d).

Fig. 3 Bright field transmission electron micrograph, Cu, C, Fe, O and combined elemental mapping images of Fe₃O₄@PAA@MOF-199, scale bar: 200 nm.

The amount of MOF-199 existing in the composite of Fe₃O₄@PAA@MOF-199 was determined by TGA. Fig. 4a shows the TGA curves of Fe₃O₄@PAA@MOF-199 (H), Fe₃O₄@PAA@MOF-199 (L) and pristine MOF-199. For the TGA curve of pristine MOF-199, there was significant mass loss between 343 °C and 369 °C, due to the decomposition of the structure. The mass loss was calculated to be about 24.2% in this temperature interval. Taking pristine MOF-199 as a reference, the mass loss of Fe₃O₄@PAA@MOF-199 (L) and Fe₃O₄@PAA@MOF-199 (H) in the same temperature interval was determined to be about 22.9% and 18.7%. Accordingly, the amount of MOF-199 in Fe₃O₄@PAA@MOF-199 (L) and Fe₃O₄@PAA@MOF-199 (H) were calibrated to be 93.4% and 77.3, respectively, by a corresponding proportional relationship. The specific surface area and porous structures of the Fe₃O₄@PAA@MOF-199s were characterized by nitrogen adsorption at 77 K. As shown in Fig. 4b, the isotherms of Fe₃O₄@PAA@MOF-199 (H) and Fe₃O₄@PAA@MOF-199 (L) are typical type I isotherms, which is the characteristic of microporous MOF-199. The hysteresis loops in the high pressure range indicate the existence of macropores, due to the packing of MOF-199 particles and some voids between Fe₃O₄@PAA submicrospheres and MOF-199 particles. The specific surface area of the samples was calculated to be 824 m² g⁻¹ and 863 m² g⁻¹ for Fe₃O₄@PAA@MOF-199 (H) and Fe₃O₄@PAA@MOF-199 (L), respectively (ESI, Table S2†). Compared with the specific surface area of pristine MOF-199 of 1286 m² g⁻¹, the decrease in specific surface area is mainly caused by the incorporation of Fe₃O₄@PAA; consequently, the lower quantities of MOF-199 in Fe₃O₄@PAA@MOF-199 (H) and the relatively lower surface area.

Fig. 4 (a) TGA curves (b) nitrogen adsorption–desorption isotherms of Fe₃O₄@PAA@MOF-199s before and after desulfurization.

Desulfurization performance

To determine an appropriate adsorption time, the adsorption profiles of DBT, BT and thiophene in model oil (5 mL) on Fe₃O₄@PAA@MOF-199 (H) (0.02 g) at 25 °C were conducted. As shown in Fig. 5, the adsorption quantity of S-compounds on Fe₃O₄@PAA@MOF-199 (H) initially increases rapidly. After that, the adsorption capacity reaches a saturated plateau and almost remained unchanged even after prolonging the adsorption time. For DBT, BT and thiophene, the saturated time was about 2.5 h, 1.5 h and 1 h, respectively, suggesting that the larger is the S-compound molecule, the longer is the saturated adsorption. Thus, 3 h was selected as an appropriate adsorption time in the following study. Fig. 6 shows the adsorption isotherms of DBT, BT and thiophene in the model oil on Fe₃O₄@PAA@MOF-199 (H) and Fe₃O₄@PAA@MOF-199 (L) and their corresponding Langmuir plots at 25 °C. The adsorption quantity of Fe₃O₄@PAA@MOF-199s increased with increasing equilibrium concentration, and Fe₃O₄@PAA@MOF-199 (L) shows higher adsorption capacity than Fe₃O₄@PAA@MOF-199 (H). The saturated adsorption capacity can be estimated by plotting C_e/q against C_e, as shown in Fig. 6, and the linear straight lines suggest that the experimental results perfectly match with the Langmuir model. As summarized in Table 1, the saturation adsorption capacity and the Langmuir constant of both Fe₃O₄@PAA@MOF-199s follow the order of DBT > BT > thiophene, which is in agreement with the electron density distribution on the S-compounds. This reveals that the adsorption capacity is mainly ascribed to the interaction between S-atom and metal ions in MOFs through π-complexation.²⁹ The saturated adsorption capacities of Fe₃O₄@PAA@MOF-199 (H) were calculated to be 35.0, 15.9 and 11.8 mg S g⁻¹ for DBT, BT and thiophene, respectively.

Fig. 5 Adsorption profiles of (a) DBT (C₀ = 371 mg L⁻¹), (b) BT (C₀ = 401 mg L⁻¹) and (c) thiophene (C₀ = 463 mg L⁻¹) in n-octane on Fe₃O₄@PAA@MOF-199 (H) at 25 °C.

Fig. 6 Adsorption isotherms of (a) DBT, (b) BT, (c) thiophene and their corresponding Langmuir model plots of (d) DBT, (e) BT, (f) thiophene on Fe₃O₄@PAA@MOF-199 (H) () and Fe₃O₄@PAA@MOF-199 (L) () after equilibrium for 3 h at 25 °C.

Table 1 Adsorption parameters of Fe₃O₄@PAA@MOF-199s based on the Langmuir model^a

Adsorbents	S-compound	Q0 (mg g^−1)	b (L mg^−1)	R	QMOF^b (mg g^−1 MOF)
a Q0 and b are the saturated adsorption capacity and Langmuir constant (L mg^−1), respectively.b QMOF adsorption capacity, calibrated to per gram of MOF-199.c Adsorption capacity of pristine MOF-199 is collected from the literature.^15
Fe3O4@PAA @MOF-199 (L)	DBT	35.0	0.00899	0.99686	37.0
	BT	15.9	0.00588	0.99897	17.0
	Thiophene	11.8	0.00389	0.99814	12.6
Fe3O4@PAA @MOF-199 (H)	DBT	25.8	0.0101	0.99967	33.4
	BT	12.4	0.00496	0.99936	16.0
	Thiophene	10.1	0.00437	0.99518	13.1
MOF-199	DBT				45^c
MOF-199	BT				25^c

---

Moreover, according to the MOF quantities in Fe₃O₄@PAA@MOF-199s determined by TGA, the adsorption capacity of S-compounds of per gram of MOF-199 on the obtained adsorbents were calibrated and compared with that on the pristine MOF-199. Take the adsorption capacity of DBT on Fe₃O₄@PAA@MOF-199 (H) as an example, it is estimated to be 37.0 mg S (g MOF)⁻¹, which coincides with the reported data.¹⁵ In addition, due to the higher quantities of MOF-199 in Fe₃O₄@PAA@MOF-199 (L), the adsorption capacity in Fe₃O₄@PAA@MOF-199 (L) is higher than Fe₃O₄@PAA@MOF-199 (H).

Recyclability and regeneration performance

Note that magnetization is the most important parameter for effective magnetic separation. Fig. 7a shows the hysteresis loop of typical magnetic MOFs measured by sweeping the external magnetic field between −1.75 and +1.75 T at the room temperature. No obvious remanence or coercivity was observed in the magnetization curve, suggesting the soft magnetic character. The saturated mass magnetization of Fe₃O₄@PAA@MOF-199 (L) and Fe₃O₄@PAA@MOF-199 (H) were estimated to be 0.3 emu g⁻¹ and 1.4 emu g⁻¹, respectively. Although the values of the saturated mass magnetization of Fe₃O₄@PAA@MOF-199s were not high, magnetic separation can still be successfully applied for separating Fe₃O₄@PAA@MOF-199s from oil (Fig. 7b). The separation performance was determined by measuring the turbidity of the oil. As shown in Table 2, the turbidity of the oil after the magnetic separation of Fe₃O₄@PAA@MOF-199 (L) and Fe₃O₄@PAA@MOF-199 (H) was 6.8 NTD and 2.1 NTD, respectively. Compared with the pristine MOF-199, after magnetic separation, its turbidity value was as high as 92.0 NTD due to no magnetic property in pristine MOF-199, suggesting the magnetic separation for Fe₃O₄@PAA@MOF-199s is significantly effective. Combining the desulfurization and magnetic separation properties, Fe₃O₄@PAA@MOF-199 (H) showed better performance than Fe₃O₄@PAA@MOF-199 (L). Therefore, incorporating a suitable amount of magnetite Fe₃O₄@PAA submicrospheres, which ensures magnetic separation efficiency without reducing the adsorption capacity, is a key point for successfully fabricating the magnetic adsorbent Fe₃O₄@PAA@MOF-199.

Fig. 7 (a) Magnetization as a function of the magnetic field for Fe₃O₄@PAA@MOF-199 (H) and Fe₃O₄@PAA@MOF-199 (L) at 25 °C. (b) Photographs of Fe₃O₄@PAA@MOF-199 (H) in n-octane before and after magnetic separation.

Table 2 Saturation magnetization and turbidity of Fe₃O₄@PAA@MOF-199s

Adsorbents	M^a (emu g^−1)	Turbidity (NTD)	DBT adsorption capacity (mg g^−1)
a M represents saturated mass magnetization obtained from VSM.
Fe3O4@PAA@MOF-199 (L)	0.3	6.8	37.0
Fe3O4@PAA@MOF-199 (H)	1.4	2.1	33.4
MOF-199		92.0

---

In real applications, the adsorbents should be renewable to reduce the operation cost. Therefore, the recycling stability is a crucial parameter. To investigate the regeneration performance, we selected p-xylene as the elution solvent to regenerate Fe₃O₄@PAA@MOF-199. After a cycle of desulfurization, magnetic separation and regeneration by p-xylene, the similar XRD patterns of Fe₃O₄@PAA@MOF-199 (H) before and after desulfurization (ESI, Fig. S4†) indicate the good stability of Fe₃O₄@PAA@MOF-199. Moreover, the TEM images of Fe₃O₄@PAA@MOF-199 (Fig. S5†) show the adsorbent still retained its original morphology in that Fe₃O₄@PAA submicrospheres were stably embedded in the MOF-199 matrix. Accordingly, we can infer that Fe₃O₄@PAA submicrospheres have a strong interaction with MOF-199 crystals, which can only be ascribed to the coordination interaction between carboxyl groups in the PAA chains and copper cations in MOF-199 crystals. Moreover, the nitrogen adsorption–desorption isotherms of Fe₃O₄@PAA@MOF-199 (L) before and after the desulfurization of DBT are compared in Fig. 4b, and the textural properties of the samples before and after the desulfurization are summarized in Table S2.† Both the specific surface area and total pore volume slightly decreased after a cycle of desulfurization and regeneration. The specific surface area of Fe₃O₄@PAA@MOF-199 (L) decreased from 863 to 792 m² g⁻¹ after the saturation adsorption of DBT. However, after regeneration by p-xylene, the specific surface area can recover to 844 m² g⁻¹. The effects of the regeneration times on the adsorption capacity are shown in Fig. 8. We can see that Fe₃O₄@PAA@MOF-199 (H) shows good adsorption stability for at least five recycles. Compared to the fresh adsorbent, the adsorption capacity decreased by less than 9% until the fifth regeneration. Therefore, the adsorbent of Fe₃O₄@PAA@MOF-199 can be easily regenerated after desulfurization and exhibit relatively stable recyclability.

Fig. 8 Regeneration performance of Fe₃O₄@PAA@MOF-199 (H) for desulfurization of DBT in n-Octane at 25 °C (C_e = 130 mg L⁻¹).

Conclusions

A novel magnetic adsorbent of Fe₃O₄@PAA@MOF-199 was prepared using a two-step assembly approach, in which PAA acted as a bridge to connect the magnetic Fe₃O₄ nanocrystals and MOF-199 crystals together. The Fe₃O₄@PAA@MOF-199 adsorbent showed good adsorption capacity for various thiophenic compounds in the model fuel, and followed the order of DBT > BT > thiophene due to the electron density distribution on the S-atom in these S-compounds. Moreover, the magnetic Fe₃O₄@PAA@MOF-199 adsorbent was beneficial in the utilization of magnetic separation. Furthermore, the Fe₃O₄@PAA@MOF-199 adsorbent exhibited stable recyclability. We expect that this novel idea of designing and fabricating magnetic MOFs could provide a potential method for adsorptive desulfurization, as well as magnetic separation.

Acknowledgements

This work is supported by the National Basic Research Program of China (2013CB733501), the National Natural Science Foundation of China (nos. 21176066, 21376074), the program of FP7-PEOPLE-2013-IRSES (PIRSES-GA-2013-612230), the 111 Project of Ministry of Education of China (no. B08021) and the Fundamental Research Funds for the Central Universities of China.

Notes and references

1. V. L. Nevel, I. Verbist, C. Harper, S. Byners, P. Smeyers, Y. Aregbe, P. Robouch, P. Taylor, G. Turk, R. Vocke and W. Kelly, IMEP report for European CommissionEUR21765EN, http://irmm.jrc.ec.europa.eu/html/interlaboratory_comparisons/imep/imep-18/EUR21765EN.pdf.
2. Z. Varga and J. Hancsok, Pet. Coal, 2003, 45, 135.
3. R. T. Yang, A. J. Hernández-Maldonado and F. H. Yang, Science, 2003, 301, 79.
4. Z. C. Li, H. P. Li, H. Zhao and G. Q. Wu, Contemp. Chem. Ind., 2013, 11, 1588.
5. V. C. Srivastava, RSC Adv., 2012, 2, 759.
6. J. Xiao, X. X. Wang, M. Fujii, Q. J. Yang and C. S. Song, AIChE J., 2013, 59, 1441.
7. A. Takahashi, R. T. Yang, C. L. Munson and D. Chinn, Langmuir, 2001, 17, 8405.
8. A. J. Hernández-Maldonado, F. H. Yang, G. Qi and R. T. Yang, Appl. Catal., B, 2005, 56, 111.
9. K. Tang, X. Hong, Y. H. Zhao and Y. G. Wang, Pet. Sci. Technol., 2011, 29, 779.
10. Y. Wang and R. T. Yang, Langmuir, 2007, 23, 3825.
11. H.-J. Jeon, C. H. Ko, S. H. Kim and J.-N. Kim, Energy Fuels, 2009, 23, 2537.
12. L. Wang, R. T. Yang and C.-L. Sun, AIChE J., 2013, 59, 29.
13. R. T. Yang, A. Takahashi and F. H. Yang, Ind. Eng. Chem. Res., 2001, 40, 6236.
14. S. Velu, X. L. Ma and C. S. Song, Ind. Eng. Chem. Res., 2003, 42, 5293.
15. K. A. Cychosz, A. G. Wong-Foy and A. J. Matzger, J. Am. Chem. Soc., 2008, 130, 6938.
16. K. A. Cychosz, A. G. Wong-Foy and A. J. Matzger, J. Am. Chem. Soc., 2009, 131, 14538.
17. H. X. Zhang, H. L. Huang, C. X. Li, H. Meng, Y. Z. Lu, C. L. Zhong, D. H. Liu and Q. Y. Yang, Ind. Eng. Chem. Res., 2012, 51, 12449.
18. N. A. Khan and S. H. Jhung, Angew. Chem., Int. Ed., 2012, 51, 1198.
19. S. S.-Y. Chui, S. M.-F. Lo, J. P. H. Charmant, A. G. Orpen and I. D. Williams, Science, 1999, 283, 1148.
20. H. Gu, K. Xu, C. Xu and B. Xu, Chem. Commun., 2006, 941.
21. M. Lewin, N. Carlesso, C.-H. Tung, X.-W. Tang, D. Cory, D. T. Scadden and R. Weissleder, Nat. Biotechnol., 2000, 18, 410.
22. S. Miltenyi, W. Müller, W. Weichel and A. Radbruch, Cytometry, 1990, 11, 231.
23. J. H. Park, G. V. Maltzahn, E. Ruoslahti, S. N. Bhatia and M. J. Sailor, Angew. Chem., Int. Ed., 2008, 120, 7394.
24. L. S. Zhong, J. S. Hu, H. P. Liang, A. M. Cao, W. G. Song and L. J. Wan, Adv. Mater., 2006, 18, 2426.
25. E. Maxwell and D. Kelland, IEEE Trans. Magn., 1978, 14, 482.
26. F. Ke, L. G. Qiu, Y. P. Yuan, X. Jiang and J. F. Zhu, J. Mater. Chem., 2012, 22, 9497.
27. M. R. Lohe, K. Gedrich, T. Freudenberg, E. Kockrick, T. Dellmann and S. Kaskel, Chem. Commun., 2011, 47, 3075.
28. J. Liang, H. Ma, W. Luo and S. Wang, Mater. Chem. Phys., 2013, 139, 383.
29. C. Yu, X. M. Fan, L. M. Yu, T. J. Bandosz, Z. B. Zhao and J. S. Qiu, Energy Fuels, 2013, 27, 1499.
30. A. R. Millward and O. M. Yaghi, J. Am. Chem. Soc., 2005, 127, 17998.

---

Footnote

† Electronic supplementary information (ESI) available: Supporting XPS measurement of Fe₃O₄@PAA, FTIR spectra of products, SEM images of Fe₃O₄@PAA@MOF-199 through ‘LBL’ method, X-ray diffraction pattern of fresh and regenerated Fe₃O₄@PAA@MOF-199 (H), TEM images of Fe₃O₄@PAA@MOF-199 after the desulfurization, copper(II) content in Fe₃O₄ and Fe₃O₄@PAA and textural parameters of Fe₃O₄@PAA@MOF-199s. See DOI: 10.1039/c4ra06515h

---