Surface molecularly imprinted polymers grafted on ordered mesoporous carbon nanospheres for fuel desulfurization

Abstract

Surface molecular imprinting is an effective measure to get adsorbent materials for definite substances. In this work, ordered mesoporous carbon nanospheres (OMCNS) were prepared as carriers by a hydrothermal method with phenolic resol as the carbon source and triblock copolymer Pluronic PF127 as the soft template. A surface molecularly imprinted polymer (SMIP/OMCNS) was then obtained through a series of functionalization, grafting and elution processes with dibenzothiophene (DBT) as a template, which was designed for the deep desulfurization of fuel products by removing DBT and its derivatives. By adopting OMCNS as a carrier, SMIP/OMCNS shows excellent adsorption capacity towards DBT (218.29 mg g⁻¹), owing to the ordered mesoporous structure and high surface area of OMCNS. The better fitted pseudo-first-order model indicates that the adsorption involves mainly physical interactions, which are promoted by the mesoporous structure of OMCNS. The Langmuir and Freundlich models fitted better than the Dubinin–Radushkevich and Scatchard isothermal models did, which suggests the occurrence of both monolayer and multilayer interactions. The thermodynamics, selectivity and regenerability of SMIP/OMCNS were also investigated. The adsorption of DBT on SMIP/OMCNS proves to be an endothermic process. And the relative selectivity coefficients against benzothiophene, biphenyl and fluorine interferents reached 2.19, 2.29 and 2.37, respectively. As a result, SMIP/OMCNS can be a potential adsorbent material for deep desulfurization of fuel products and recovering DBT and its derivatives as valuable reagents for other value-added applications.

---

1. Introduction

New environmental regulations regarding the sulfur content in transportation fuels have forced researchers and refineries to develop more efficient processes for producing cleaner fuels than ever.^1,2 Dibenzothiophene (DBT) and its derivatives are widely distributed in transportation fuels and account for more than 60% of the total sulfur, for which the conventional hydrodesulfurization (HDS) process is somewhat inefficient. On the other hand, DBT and its derivatives are also valuable intermediates in the production of medicines, pesticides, thioindigo dyes, liquid crystal polymers and optoelectronic materials,³ which are however destructively processed every year. It is then urgent to find a way to deeply, selectively and non-destructively remove DBT and its derivatives with high efficiency from fuels and turn them into valuable chemical reagents.

Surface molecular imprinting is an effective measure to get adsorbent materials selective for definite substance. With this measure, the surface molecularly imprinted polymer (SMIP) layer can be grafted on certain carriers and targeted for deep desulfurization.⁴ Because the specific molecular recognition sites are created during the grafting process, SMIP, as an advanced adsorbent, is then provided with unique advantages in selective adsorbing and removal of specific organic sulfur-containing compounds from fuels in the presence of other similar interferential compounds.^5,6

The adsorption desulfuration with SMIP is considered as an attractive approach in comparison with other processes such as oxidation,^7,8 HDS,^9,10 extraction^11,12 and biodesulfurization,^13,14 owing to the easy operation and low cost. Especially, SMIP is effective for deep desulfurization without any detriment to the fuel quality, as it does not break the bonds and structures of the olefins and aromatic hydrocarbons in fuels besides the target DBT molecules.

By locating the imprinted sites on the carrier surface, SMIP is advantageous for easy elution, high selectivity, good accessibility and fast mass transfer, in comparison with traditional bulky imprinted materials. It is generally difficult to get bulky imprinted polymers in uniform particles¹⁵ and their rigid structure also makes the extraction or adsorption of templates difficult through the thick cross-linked polymeric layer.¹⁶

The nature of a carrier is crucial to get an SMIP with demanded properties such as mono-dispersion, homogeneity, uniform orientation and excellent mechanical, chemical and thermal stabilities. Nowadays, multitudinous materials including Al₂O₃,⁹ TiO₂,^17,18 K₂Ti₆O₁₃,¹⁹ CdS, ZnS quantum dots, SiO₂,²⁰ silica gel,²¹ carbon-microspheres,^22,23 carbon nanotubes^24,25 and graphene²⁶ have been used as the carriers for SMIP. Among them, porous carbon materials, especially ordered mesoporous carbon nanosphere (OMCNS), are supposed to be ideal carrier for preparing SMIP, because of their high porosity, high surface area, low density, high mechanical stability, large pore volume, easy functionalization, high mechanical and thermal stability, as well as good biocompatibility.^27,28 The open-framework, ordered mesoporous structure and nanosized particles of OMCNS make it widely applicable in adsorption and separation,²⁹ controllable drugs release,³⁰ electrochemical energy storage,³¹ and catalysis.³²

In this work, OMCNS was prepared by hydrothermal method with phenolic resol as carbon source and triblock copolymer Pluronic PF127 as soft template. Designed for the deep desulfurization of fuel products by removing DBT and its derivatives and collecting them as valuable chemicals, the corresponding surface molecularly imprinted polymer (SMIP/OMCNS) was then obtained through a series of functionalization, grafting and elution processes with OMCNS as carrier and dibenzothiophene (DBT) as template. The adsorption kinetics, thermodynamics, selectivity and regenerability of SMIP/OMCNS were investigated by gas chromatography (GC). The synergistic effect between SMIP and OMCNS on the excellent performance of SMIP/OMCNS in selective adsorption of DBT was then elucidated.

2. Methods

2.1. SMIP/OMCNS preparation

Preparation of OMCNS carrier. As depicted in Fig. 1, OMCNS was prepared through a hydrothermal synthesis procedure.²⁷ First, 0.6 g of phenol, 2.1 mL of formalin aqueous solution (37 wt%) and 15 mL of NaOH aqueous solution (0.1 mol L⁻¹) were mixed into a three-neck flask. The mixture solution was stirred at 70 °C for 0.5 h to obtain a low-molecular-weight phenolic resol. Then, 0.96 g of triblock copolymer Pluronic F127 (M_w = 12 600, PEO₁₀₆PPO₇₀PEO₁₀₆, SIGMA-Aldrich) dissolved in 15 mL of H₂O was added and the mixture was stirred at 66 °C for 2 h, followed by the addition of 50 mL of H₂O to dilute the solution, leaving the mixture reaction for another 17 h. After that, the reaction kettle filled with 10 mL of the resultant solution and 30 mL of H₂O was heated at 130 °C for 24 h. The products were collected after centrifugation and washed with deionized water for several times and dried at room temperature. At last, OMCNS was obtained by carbonization and removal of PF127 at 700 °C in nitrogen atmosphere for 3 h.

Fig. 1 Schematic procedures for the preparation of OMCNS and SMIP/OMCNS.

Preparation of SMIP/OMCNS. The special identifiable adsorption property of SMIP comes from their particularly designed structures. Several steps are indispensable during the preparation of SMIP on OMCNS. γ-Methacryloxypropyl trimethoxysilane (KH570) is a commonly used amphiphilic agent in the preparation of inorganic/organic complex.^20,26 Thus, KH570 was used to modify the surface chemical conditions of OMCNS by introducing –C=C– bridge for subsequent MAA grafting. Typically, 0.3 g of OMCNS, 1 mL of KH570 and 60 mL of mixed solvent (C₂H₅OH/H₂O, 3 : 1 by volume) were added into a three-neck flask; the pH value was modified to 5 by acetic acid. The mixture was stirred at 65 °C for 2 h and the products were filtered and washed with ethanol, and then dried at 50 °C overnight, to get KH570 grafted OMCNS (KH570/OMCNS).

After that, 0.2 g of KH570/OMCNS, 20 mL of H₂O, 0.105 g of (NH₄)₂S₂O₈ and 1 mL of functional monomer methacrylic acid (MAA, 99%) were sufficiently mixed in a three-neck flask and stirred under nitrogen atmosphere at 80 °C for 24 h. The products were filtered and washed with ethanol, and then dried at 50 °C overnight, to obtain polymethacrylic acid (PMAA) grafted KH570/OMCNS (PMAA/OMCNS). Grafting of functional monomer is the key step of SMIP preparation. Since certain intermolecular interactions exist between functional monomer and template molecule, the adsorption of SMIP largely depends on the template–monomer complex interactions. For the template molecule DBT, MAA is a favourable monomer according to our previous work.^33,34 The complex DBT–MAA processes a certain degree of binding energy to guarantee the ability of corresponding SMIPs for the identification and adsorption of DBT.

After MAA grafting and template pre-assembling, the crosslinking agent EDMA is needed to form a reticularly structured polymer layer to fix the DBT–MAA complex sites. Thus 0.1 g of PMAA/OMCNS, 0.1843 g of DBT and 10 mL of chloroform were sufficiently mixed and stirred for 0.5 h at 25 °C to pre-assemble DBT with PMAA/OMCNS. Then 4 mL of ethylene glycol dimethacrylate (EDMA, 98%) as cross-linking agent was added under continuous stirring; the mixture was heated to 50 °C and held at this temperature for another 10 h.

Finally, the product was washed successively with methanol/acetic acid (9 : 1, by volume) to remove DBT, until no DBT could be detected in the eluent by GC analysis. Finally, SMIP/OMCNS was obtained by drying the collected product overnight at 50 °C.

For comparison, the non-imprinted polymer carried on OMCNS, denoted as NIP/OMCNS, was prepared via the same procedures without adding DBT as template.

2.2. Characterization

The morphologies and structures of adsorbent products were characterized by field emission scanning electron microscopy (FESEM; JSM-6700F, operated at 10 kV), transmission electron microscopy (TEM; JEM-100CXII, acceleration voltage of 100 kV, dot resolution 0.3 nm). The textural properties were characterized by an automated surface area and pore size analyzer (Quadrasorb SI). Fourier transform infrared (FT-IR) spectra were measured on a Bruker Tensor 27 FT-IR spectrometer. Thermogravimetric (TG) profiles were measured on a Netzsch TG 209 F3 thermogravimeter; the sample was heated from 40 to 900 °C in nitrogen at a heating rate of 10 °C min⁻¹.

2.3. Static adsorption

In this study, the GC (GC-2014C, Shimadzu, Japan) was used to measure the concentration of DBT in n-hexane. The capillary chromatographic column (Rtx-1: 30 m, 0.25 mm, 0.25 μm, 330/350 °C) and flame ionization detector (FID: <400 °C, >3 pg C s⁻¹) are adopted in this GC instrument. The temperatures of detector, vaporizer and column were set at 300, 300 and 220 °C.

Adsorption kinetics. A model solution was prepared by dissolving DBT in n-hexane with an initial concentration C₀ of 3.0 mmol L⁻¹. SMIP/OMCNS, NIP/OMCNS or OMCNS (50 mg) was put into a conical beaker with 100 mL of the model solution. The beaker was placed in a water bath with magnetic stirring at a fixed temperature (298, 308 or 318 K). The mixture was then sampled with a 1 mL syringe at designed intervals (t = 5, 10, 15, 25, 35, 45, 60, 75, 90, 110, 130 and 150 min) until the concentration of DBT was settled. Then the syringe filters were used to remove the adsorbents dispersed in the solution. The clear DBT solutions were injected into the sampling bottles. After that, the liquid samples were analyzed by GC to get the residual DBT concentration (C_t, mmol L⁻¹).

According to values of C₀ and C_t, the adsorption capacities (Q, mg g⁻¹) at given times (t) can be calculated by eqn (1):

Q = MV(C₀ − C_t)/m (1)

where M (g mol⁻¹) is the molecular mass of DBT adsorbate, V (L) is the volume of DBT n-hexane solution, and m (g) is the weight of the adsorbent.

Adsorption isotherms. A series of SMIP/OMCNS adsorbent samples (5 mg each) was introduced into different conical beakers containing 10 mL of the DBT n-hexane solution with various initial concentrations (C₀: 1.0, 2.0, 3.0, 4.0 and 5.0 mmol L⁻¹). According to the kinetic curves, it can be found that the equilibrium times of SMIP/OMCNS towards DBT at 298, 308 and 318 K are around 1.5 h. Therefore, we have carried through the isothermal experiments at a fixed temperature (298, 308 and 318 K) for 2 h to ensure the adsorption equilibriums can be reached. The saturated SMIP/OMCNS was obtained after reaching the adsorption equilibrium. At the same time, the mixtures were drawn out with syringes and the equilibrium concentrations (C_e, mmol L⁻¹) were determined by GC. The equilibrium adsorption capacity (Q_e, mg g⁻¹) is obtained by

Q_e = MV(C₀ − C_e)/m (2)

2.4. Selective adsorption

The adsorption selectivity of SMIP/OMCNS towards DBT was investigated against BT, biphenyl and fluorine as interfering molecules that have similar structures with DBT. The concentration of each substance in the mixed solution of DBT with BT/biphenyl/fluorine was all kept as 3.0 mmol L⁻¹. Typically 50 mg of the adsorbent (SMIP/OMCNS or NIP/OMCNS) was added and dispersed in 100 mL of the mixed solution. After that, the mixture was placed in a water bath at 298 K for 3 h for an equilibrium adsorption; the concentrations of each ingredient in the supernatant were then determined by GC.

2.5. Regeneration of spent SMIP/OMCNS

The SMIP/OMCNS adsorbent (50 mg) saturated with DBT after the static adsorption test was collected by centrifugation and washed with a mixture of methanol/acetic acid (9 : 1, by volume) with a suction filter to completely elute DBT. The regenerated SMIP/OMCNS was obtained by suction filtering and washing thoroughly with ethanol and H₂O after completely removing residual methanol and acetic acid. The changes of DBT concentrations in filtrate with the dosage of methanol/acetic acid were analyzed by GC. In addition, the recovered SMIP/OMCNS adsorbent was then reused for the subsequent adsorption tests. To test the recyclability of SMIP/OMCNS adsorbent, the adsorption–desorption cycles were repeated for 6 cycles under the same conditions.

3. Results and discussion

3.1. Morphologies and porous structures of OMCNS and SMIP/OMCNS

As shown by the FESEM and TEM images in Fig. 2, OMCNS is in spheres with a crude surface and displays a nice dispersibility and uniformity with a diameter (d_sphere) around 130 nm. Though influenced by the direction of observation and the interference of carbon film substrate layer, the ordered porous structure of OMCNS can be observed from TEM images (inset in Fig. 2(d)).

Fig. 2 FESEM images of (a) OMCNS and (c) SMIP/OMCNS; TEM images of (b) OMCNS and (d) SMIP/OMCNS.

In comparison with OMCNS, however, SMIP/OMCNS exhibits rougher surfaces; larger particles and certain degree of agglomeration owing to the coating of imprinted polymer layer. The sizes of OMCNS and SMIP/OCNS are statistically analyzed as 139.02 and 156.89 nm, respectively from the FESEM images according to Fig. 3(a). The diameter increased 17.87 nm from OMCNS to SMIP/OMCNS. And the overall thickness of the polymer layer is 8.94 around. Besides, the ordered porous structure can be clearly observed in the TEM images of SMIP/OMCNS. And the size distributions of OMCNS and SMIP/OMCNS indicate that the average pore sizes of OMCNS and SMIP/OMCNS are 3.08 and 2.01 nm, respectively according to Fig. 3(b).

Fig. 3 (a) Particle size distributions of OMCNS and SMIP/OMCNS and (b) pore size distributions of OMCNS, SMIP/OMCNS and regenerated SMIP/OMCNS.

The textural properties of OMCNS, SMIP/OMCNS, saturated SMIP/OMCNS and regenerated SMIP/OMCNS are summarized in Table 1. After imprinting, SMIP/OMCNS has a lower surface area (327.61 m² g⁻¹) than the OMCNS carrier (488.02 m² g⁻¹). Meanwhile, the total pore volume and average pore size of SMIP/OMCNS are also decreased after imprinting. The coating of imprinted polymer layer leads to a moderate increase in particle size, but a decrease in specific surface area and pore volume and diameter, indicating that the mesoporous structure of OMCNS carrier has been effectively used in the preparation of SMIP/OMCNS. However, the mesoporous structures of OMCNS would prevent the whole surface of OMCNS to be covered with polymer layer completely. The thicknesses of polymer layer could not be all the same on all parts of SMIP/OMCNS since it can be speculated that the thickness of polymer layer on the inner walls of pores could be thinner than that on the surrounding areas of pores. Thus, the surface area of OMCNS still plays a role in dictating the overall surface area of SMIP/OMCNS.

Table 1 Specific surface area and porous parameters of OMCNS, SMIP/OMCNS, saturated SMIP/OMCNS and regenerated SMIP/OMCNS^a

Adsorbent	S (m^2 g^−1)	Vpore (cm^3 g^−1)	dpore (nm)
a Note: The surface area (S) was calculated by the multipoint BET model from nitrogen adsorption data; the pore volume (Vpore) was calculated by the V–t plot method; and the pore diameter (dpore) was determined by the BJH model from the desorption branches of the isotherms. OMCNS was preprocessed at 300 °C while SMIP/OMCNS, saturated SMIP/OMCNS and regenerated SMIP/OMCNS were preprocessed at 100 °C by considering their thermal stabilities.
OMCNS	488.02	0.193	3.08
SMIP/OMCNS	327.61	0.079	2.01
Saturated SMIP/OMCNS	290.32	0.041	2.00
Regenerated SMIP/OMCNS	390.58	0.068	2.01

---

Moreover, after fully adsorbing DBT, the saturated SMIP/OMCNS shows even smaller surface area, total pore volume and average pore size since the imprinting cavities are loaded with DBT. On the other hand, according to the data of regenerated SMIP/OMCNS, the pore structures can be well recovered by removing the DBT with methanol/acetic acid (9 : 1, by volume) solution.

The FT-IR spectra of OMCNS, SMIP/OMCNS and the intermediates during molecular imprinting are shown in Fig. 4(a). A large amount of oxygen-containing functional groups are distributed on their surface. The bands at 3435 and 1631 cm⁻¹ are ascribed to –OH and –C=O stretching vibration, respectively. From OMCNS to SMIP/OMCNS, no new bands are found, indicating no new functional groups are introduced. It should be noted that KH570/OMCNS exhibits much stronger bands for the oxygen-containing functional groups than OMCNS owing to the grafting of KH570.³⁵ As a result of the graft copolymerization of MAA, however, these absorption bands become weaker for PMMM/OMCNS, SMIP/OMCNS, and NIP/OMCNS.

Fig. 4 (a) FT-IR spectra of the adsorbent products at different steps; (b) TG curves of the adsorbent products at different steps (N₂, 10 K min⁻¹).

The TG curves of OMCNS, SMIP/OMCNS and the intermediates during preparation upon heating in N₂ from 100 to 800 °C are illustrated in Fig. 4(b) for the purpose of finding the amount of each reagent (KH570, MAA and EDMA) that is grafted on. After annealing, OMCNS exhibits excellent thermal stability with a weight loss of 1.5% till 800 °C. The total weight loss of KH570/OMCNS is around 4.6% till 800 °C, which is caused by the decomposition of KH570 at 350–500 °C. And the total weight loss of PMAA/OMCNS is around 8.9% till 800 °C, which is caused by the weight losing of KH570 and PMAA at 350–500 °C and 150–350 °C, respectively. Correspondingly, the weight loss of SMIP/OMCNS upon heating till 800 °C is 12.1%, a little less than that of NIP/OMCNS (13.3%).

PMAA grafting is the most significant step for the preparation of SMIP/OMCNS, through which the functional groups for DBT recognizing and capturing are introduced. As a result, the degree of PMAA grafting (D_g) is an indicator for the adsorption ability towards DBT,²² which is defined as

D_g = (W_later − W_former)/W_former × 100% (3)

where W_former and W_later are the weight losses of adsorbent products before and after the PMAA grafting, respectively. As shown in Fig. 4(b), the PMAA grafting degree on OMCNS reaches 93.5%; such a high grafting degree evinces that abundant recognition sites have been introduced, owing to the large surface area and ordered mesoporous structure of OMCNS.

3.2. Adsorption kinetics of DBT on SMIP/OMCNS

The kinetic adsorption curves of DBT on SMIP/OMCNS, NIP/OMCNS and OMCNS at various temperatures are shown in Fig. 5. All the adsorption profiles are in a similar tendency; the adsorption proceeds very fast at the beginning and then levels off gradually after adsorption for 30 min, reaching an equilibrium after around 1.5 h. The change in the adsorption rate comes from the gradually occupied pores or imprinted cavities on the adsorbent surface; with the filling of the pores and imprinted cavities, the diffusion and adsorption of DBT molecules become more difficult. The saturated adsorption capacity of SMIP/OMCNS towards DBT reaches 218.29 mg g⁻¹ at 298 K, which is outstanding in comparison with those reported in other previous works.^17–26

Fig. 5 Kinetic adsorption curves of DBT dissolved in n-hexane solution on SMIP/OMCNS, NIP/OMCNS and OMCNS.

As expected, SMIP/OMCNS exhibits a much higher saturated DBT adsorption capacity (218.29 mg g⁻¹) than both NIP/OMCNS (95.07 mg g⁻¹) and OMCNS (68.22 mg g⁻¹) at 298 K. SMIP/OMCNS and NIP/OMCNS are covered with polymer layers on their surface, which may improve their adsorption abilities. However, only SMIP/OMCNS is provided with imprinted polymer layer with specially designed recognition sites and cavities, which brings the specific binding with DBT molecule by electrostatic interaction, hydrogen bonds and shape selection. In comparison, the polymer layer on NIP/OMCNS does not have suitable imprinted cavities and recognition sites, giving NIP/OMCNS a much lower adsorption capacity towards DBT. The adsorption superiority of SMIP/OMCNS over NIP/OMCNS can be intuitively described with an imprint factor (f_imp),³⁶

f_imp = Q_e(SMIP)/Q_e(NIP) (4)

where Q_e(SMIP) and Q_e(NIP) are the saturated adsorption amounts (mg g⁻¹) of SMIP/OMCNS and NIP/OMCNS towards DBT at the same temperature, respectively. The f_imp value of SMIP/OMCNS over NIP/OMCNS is 2.30, indicting a very effective imprinting for SMIP/OMCNS in this work.

Furthermore, the adsorption profiles are also fitted with pseudo-first-order (eqn (5)) and pseudo-second-order (eqn (6)) kinetic models,^36,37

ln(Q_e − Q_t) = ln Q_e − k₁t/2.323 (5)

t/Q_t = 1/(k₂Q_e²) + t/Q_e (6)

where Q_e and Q_t (mg g⁻¹) are the amounts of DBT adsorbed on SMIP/OMCNS or NIP/OMCNS at equilibrium and time t (min), respectively; k₁ (min⁻¹) and k₂ (g mg⁻¹ min⁻¹) are the pseudo-first-order and pseudo-second-order adsorption rate constants, respectively.

The fitting results are summarized in Table 2. By viewing from the value of R², the pseudo-first-order model fits better than the pseudo-second-order one. It is then speculated that the adsorption involves mainly physical interaction between DBT and the adsorbent. Besides, there are no new chemical bonds formed or destructed during the adsorption process and the adsorption rate and equilibrium time are not significantly affected by the temperature, which also suggests that the adsorption process of DBT on the surface of SMIP/OMCNS can be judged as a physical adsorption process. However, the excellent adsorption ability of SMIP/OMCNS beyond that of OMCNS comes from the imprinted polymer layer, on which are distributed many imprinted cavities that are tailored for DBT. The size and shape of cavities are matching with DBT. Besides, the functional monomers in these cavities provide some non-covalent interactions with DBT, such as ionic bonds, hydrogen bonds, electrostatic interaction and van der Waals force. Although the binding energies of non-covalent bonds are relatively small, multiple non-covalent bonds and multiple interaction sites working together can be strong enough to interact with DBT.^5,33 Along with the shape selectivity of the cavities towards DBT SMIP/OMCNS process much more excellent identity adsorption capacity than OMCNS does.

Table 2 Kinetic constant for the adsorption of DBT dissolved in n-hexane solution onto SMIP/OMCNS and NIP/OMCNS

Adsorbent	T (K)	Qe,exp (mg g^−1)	Pseudo-first-order			Pseudo-second-order
			Qe,cal (mg g^−1)	k1 (min^−1)	R^2	Qe,cal (mg g^−1)	k2 (g mg^−1 min^−1)	R^2
NIP/OMCNS	298	95.07	94.63	0.1225	0.9913	98.70	0.00187	0.9889
SMIP/OMCNS	298	218.29	216.78	0.0856	0.9970	232.75	0.00041	0.9896
	308	227.55	225.62	0.0905	0.9972	239.52	0.00043	0.9900
	318	238.98	237.49	0.0983	0.9976	252.09	0.00045	0.9872

---

3.3. Adsorption isotherm of DBT on SMIP/OMCNS

The saturation adsorption capacities of SMIP/OMCNS towards DBT at a series of DBT concentrations (1.0, 2.0, 3.0, 4.0, 5.0 mmol L⁻¹) and temperatures (298, 308 and 318 K) were determined and fitted with Langmuir (eqn (7)), Freundlich (eqn (8)) Dubinin–Radushkevich (eqn (9)) and Scatchard (eqn (10)) equations,^19,38,39

C_e/Q_e = C_e/Q_m + 1/(Q_mk_L) (7)

ln Q_e = ln C_e/n + ln k_F (8)

ln Q_e = ln Q_m − βε² (9)

Q_e/C_e = k_Sb − Q_eb (10)

ε = RT ln(1 + 1/C_e) (11)

where Q_e (mg g⁻¹) and C_e (mmol L⁻¹) are adsorption amount and concentration of DBT at adsorption equilibrium, respectively; Q_m (mg g⁻¹) represents maximum adsorption capacity; k_L (L mmol ⁻¹) is the Langmuir constant, which is related to the affinity of binding sites; k_F and n are both Freundlich constants; β (mol² kJ⁻²) is the coefficient related to the mean free energy of adsorption; ε is Polanyi potential that can be obtained according to eqn (11); k_S (mg g⁻¹) and b (mmol L⁻¹) are both Scatchard adsorption isotherm parameters; R is the ideal gas constant; T (K) is the absolute temperature.

The nonlinear regressions of Langmuir and Freundlich isotherm models for the adsorption of DBT on SMIP/OMCNS are illustrated in Fig. 6 since these two models fit better with the experimental results. All the isotherm curves at various temperatures show similar tendency; the adsorption amount increases rapidly with the DBT concentration at low concentration and levels off at high concentration.

Fig. 6 Adsorption isotherms of DBT dissolved in n-hexane solution on SMIP/OMCNS at different temperatures.

As summarized in Table 3, the values of Freundlich constant n are all larger than 1, representing a favourable adsorption. By viewing from the correlation coefficient R², both Langmuir and Freundlich models are fairly good in fitting the adsorption data. In fact, both models have some limitations in describing the adsorption processes on SMIP/OMCNS over a wide range of DBT concentration, even though they are widely used.²³ Langmuir model is used to describe the monolayer adsorption with only one kind of binding site, whereas Freundlich equation is usually used to describe the multilayer adsorption and thus fits some poorly near the saturated portion. The Dubinin–Radushkevich isotherm is used to estimate the porosity of adsorbent. The lower R² values for Dubinin–Radushkevich fittings can be ascribed to the porous adsorption occurred on SMIP/OMCNS. At the same time, the Scatchard plot, known as independent site-oriented model, helps to evaluate the adsorption. The lower R² values for Scatchard model fitting indicate a heterogeneous surface of SMIP/OMCNS and various types of active sites involved in the adsorption process. Taking all these factors into consideration, leads to the speculation that the adsorption of DBT on SMIP/OMCNS may be an integration of monolayer and multilayer adsorption, as the imprinted layer generally presents certain heterogeneity. Some porous adsorption also takes place during the adsorption process.

Table 3 Parameters of Langmuir and Freundlich isothermal adsorption equations fitted for the adsorption data of DBT on SMIP/OMCNS

T (K)	Langmuir			Freundlich
	Qm (mg g^−1)	kL (L mmol^−1)	R^2	n	kF	R^2
298	231.48	7.794	0.9702	6.238	191.4	0.9781
308	237.53	14.403	0.9731	8.077	208.4	0.9718
318	250.63	15.465	0.9751	7.830	220.3	0.9708

---

T (K)	Dubinin–Radushkevich			Scatchard
	Qm (mg g^−1)	β (mol^2 kJ^−2)	R^2	kS (mg g^−1)	b (mmol L^−1)	R^2
298	223.60	0.0237	0.9336	234.08	7.253	0.9401
308	255.12	0.0134	0.9519	238.89	13.624	0.9484
318	246.29	0.0120	0.9557	252.03	14.628	0.9483

---

3.4. Adsorption thermodynamics of DBT on SMIP/OMCNS

Another character which can be got from the adsorption isotherms (Fig. 6) is that the adsorption capacity of SMIP/OMCNS towards DBT increases with the rise of temperature, suggesting an endothermic process.⁴⁰ The thermodynamic parameters including standard free energy change ΔG^θ, standard enthalpy change ΔH^θ, and standard entropy change ΔS^θ can be obtained according to eqn (12)–(14).³⁸

ΔG^θ = −RT ln K_c (12)

ln K_c = −ΔH^θ/(RT) + const (13)

ΔS^θ = (ΔH^θ − ΔG^θ)/T (14)

where K_c is equilibrium constant (K_c = Q_e/C_e, mg L g⁻¹ mmol⁻¹).

As given in Table 4, the negative value of ΔG^θ and positive value of ΔS^θ suggest that the adsorption of DBT on SMIP/OMCNS takes place spontaneously. The absolute value of ΔG^θ is lower than 20, which confirms that the overall adsorption process is dominated by physical interaction, in accordant with the adsorption kinetic analysis. Moreover, the adsorption is an endothermic process with a positive ΔH^θ, in accordant with above adsorption isotherm analysis.

Table 4 Thermodynamic parameters for the adsorption DBT in n-hexane on SMIP/OMCNS

T (K)	Kc (mg L g^−1 mmol^−1)	ΔG^θ (kJ mol^−1)	ΔH^θ (kJ mol^−1)	ΔS^θ (kJ mol^−1 K^−1)
298	159.60	−12.57		0.0422
308	177.19	−13.26	7.39	0.0431
318	184.37	−13.79		0.0434

---

The endothermic feature can also be considered as an evidence to prove that physical interactions predominate in the overall adsorption process, because previous theoretical calculations pointed out that the chemical interaction between DBT and functional monomer MAA is exothermic. Like most solid–solute interface adsorptions ever reported, the adsorption capacity can be promoted by rising the temperature.^41,42 On one hand, the mass transfer, pore diffusion and orientation adjustment of DBT towards the imprinted sites can be accelerated by rising temperature; on the other hand, the solvent influence is weakened under real adsorption conditions.⁴³ As a whole, the adsorption of DBT on SMIP/OMCNS presents as an endothermic process.

3.5. Adsorption selectivity of SMIP/OMCNS towards DBT against other interferents

One of the unique advantages of SMIP materials is their specific recognition ability towards target molecules in the presence of certain interferents. Only those molecules that have specific size and conformational structure as well as certain intermolecular affinity with the imprinted cavities can be well recognized and trapped. The identification ability of SMIP/OMCNS towards DBT was tested in the presence of interferents such as BT, biphenyl and fluorine, as shown in Fig. 7.

Fig. 7 Adsorption selectivity of SMIP/OMCNS towards DBT against BT, biphenyl and fluorene interferents.

The adsorption capacity of SMIP/OMCNS towards DBT is much larger than that towards other similar molecules, because the imprinted cavities are just tailor-made for DBT and do not match with other three analogical interferents in shape, size and spatial arrangement.^22,26 Noteworthily, SMIP/OMCNS exhibits much higher adsorption capacity and better selectivity towards DBT than NIP/OMCNS, as the later does not have the DBT imprinted polymer layer. As to OMCNS, it shows no selectivity towards DBT in the same mixed solution. The adsorption amounts towards DBT, BT, biphenyl and fluorine are almost same since only nonselective porous adsorption occurs on the OMCNS. Besides, the lack of loosen polymer layer on the surface of OMCNS causes the relatively low adsorption amounts of each molecule.

Three interferents are also different in their molecular structure and interaction with the adsorbent. BT molecule with a smaller size may enter and leave the cavities more easily, but its mismatched spatial shape and higher energy gap give it less opportunity to be sited in the DBT imprinted cavities.⁴⁴ On the other hand, although biphenyl and fluorene are dimensionally close to DBT, the lack of specific interaction sites and relatively larger steric hindrance also make them harder to be adsorbed on SMIP/OMCNS. All these results suggest that SMIP/OMCNS is able to identify and adsorb DBT selectively; the selective adsorption ability can be described by the relative selectivity coefficient (K and K′), following eqn (15)–(17),^17,23

K_d = Q_e/(MC_e) (15)

K = K_d(DBT)/K_{d(interferent)} (16)

K′ = K_(SMIP)/K_(NIP) (17)

where K_d is the distribution coefficient (L g⁻¹); K and K′ are selectivity coefficients for given adsorbate and adsorbent, respectively.

As given in Table 5, the relative selectivity coefficients (K′) of SMIP/OMCNS towards DBT against BT, biphenyl and fluorene are 2.19, 2.29 and 2.37, respectively. It illustrates that SMIP/OMCNS has a stronger adsorption affinity and selectivity towards DBT with respect to NIP/OMCNS, owing to the matched binding sites and cavity size with DBT in SMIP/OMCNS. Meanwhile, the distribution coefficient (K_d) of SMIP/OMCNS towards DBT is much greater than those towards the BT, biphenyl or fluorene interferents.

Table 5 Adsorption selectivity of SMIP/OMCNS towards DBT against the BT, biphenyl and fluorene interferents

Adsorbate	SMIP/OMCNS			NIP/OMCNS			K′
	Qe (mg g^−1)	Kd (L g^−1)	K	Qe (mg g^−1)	Kd (L g^−1)	K
DBT	131.24	0.3113	—	62.35	0.1271	—	—
BT	54.60	0.1569	1.98	49.48	0.1401	0.91	2.19
Biphenyl	61.72	0.1540	2.02	58.24	0.1440	0.88	2.29
Fluorene	58.33	0.1325	2.35	56.73	0.1284	0.99	2.37

---

3.6. Recyclability of SMIP/OMCNS for DBT adsorption

The stability and recyclability of an adsorbent are two key factors for any practical application. According to Table 1 and Fig. 3(b), the surface area (S, m² g⁻¹), total pore volume (V_pore, cm³ g⁻¹) and average pore size (d_pore, nm) of the saturatedly adsorbed SMIP/OMCNS become smaller than those of SMIP/OMCNS since DBT has been adsorbed on the surface of SMIP/OMCNS. After the regeneration experiment, the DBT has been removed from the saturatedly adsorbed SMIP/OMCNS by washing with the methanol/acetic acid (9 : 1, by volume). The regenerated SMIP/OMCNS processes higher values of S, V_pore and d_pore than saturatedly adsorbed SMIP/OMCNS. Furthermore, these values are similar with corresponding values of as-prepared SMIP/OMCNS, which indicates the good stability and recyclability of SMIP/OMCNS.

According to the adsorption experiments, the saturated adsorption capacity of SMIP/OMCNS is 218.29 mg g⁻¹. Thus, the total content of DBT on 50 mg of saturatedly adsorbed SMIP/OMCNS is around 10.91 mg. The changes of DBT concentrations in filtrate with the dosage of methanol/acetic acid were analyzed by GC as shown in Fig. 8(a), the amount of DBT in 100 mL of filtrate is 7.49 mg and the value increases with the dosage of methanol/acetic acid. The amount of DBT in 1000 mL of filtrate reaches 9.93 mg, which means that 91% of DBT can be recovered from SMIP/OMCNS.

Fig. 8 (a) Recovery efficiency of DBT in methanol/acetic acid mixture and (b) the recyclability of SMIP/OMCNS for DBT adsorption.

The spent SMIP/OMCNS adsorbent can be effectively regenerated through washing with methanol/acetic acid. As illustrated in Fig. 8(b), the adsorption capacity of SMIP/OMCNS towards DBT declines only about 12.53% after six cycles. The slight drop in adsorption ability may be attributed to the low density and strength of mesoporous OMCNS. However, the binding and release ability of SMIP/OMCNS for DBT remains almost unchanged, meaning that the identification ability and specific memory effect towards DBT are maintained. As a result, SMIP/OMCNS can be considered as a regenerable desulfurization adsorbent with excellent structural stability.

4. Conclusion

OMCNS with ordered mesoporous structure was prepared as carrier through soft-template hydrothermal method, which has a particle size of 139.02 nm, a surface area of 488.02 m² g⁻¹, and pore volume of 0.193 m³ g⁻¹ with an average pore size of 3.08 nm. With OMCNS as a carrier, SMIP/OMCNS adsorbent was then obtained with DBT as template. In comparison with OMCNS, after grafting with SMIP, the particle size of SMIP/OMCNS is increased by 8.94 nm; the surface area, pore volume and average pore size are decreased to 327.61 m² g⁻¹, 0.079 cm³ g⁻¹ and 2.01 nm, respectively. FT-IR and TG analysis results illustrate that the imprinted polymer layer is grafted on the surface of mesoporous OMCNS.

The adsorption kinetics, thermodynamic, selectivity and recyclability of the SMIP/OMCNS adsorbent towards DBT were then investigated. The results demonstrate that SMIP/OMCNS is provided with outstanding adsorption capacity towards DBT (218.29 mg g⁻¹, 298 K), in comparison with those previously reported. The adsorption of DBT on SMIP/OMCNS is an endothermic process involving monolayer and multilayer physical interaction. Some porous adsorption also takes place during the adsorption process. Moreover, SMIP/OMCNS is able to recognize DBT selectively for adsorption against the interferents with similar structures; the relative selectivity coefficients of SMIP/OMCNS towards DBT against BT, biphenyl and fluorene reach 2.19, 2.29 and 2.37, respectively. The spent SMIP/OMCNS adsorbent can be effectively regenerated through washing with methanol/acetic acid; the adsorption capacity of SMIP/OMCNS towards DBT declines only slightly after six adsorption–regeneration cycles. All these suggest that SMIP/OMCNS obtained in this work can be a potential adsorbent material for deep desulfurization of fuel products and recovering of DBT and its derivatives as valuable reagents for chemical synthesis.

Acknowledgements

This work was financially supported by National Natural Science Foundation of China (No. 21176169), International Science & Technology Cooperation Program of China (No. 2012DFR50460) and Shanxi Provincial Key Innovative Research Team in Science and Technology (No. 2015013002-10).

References

1. Y. S. Shen, P. W. Li, X. H. Xu and H. Liu, RSC Adv., 2012, 2, 1700.
2. A. S. Ogunlaja, C. D. Sautoy, N. Torto and Z. R. Tshentu, Talanta, 2014, 126, 61.
3. N. Uramaru, E. C. Chang, W. P. Yen, M. Y. Yeh, S. H. Juang and F. F. Wong, Arabian J. Chem., 2015 DOI:10.1016/j.arabjc.2014.12.004.
4. Y. Z. Yang, X. G. Liu and B. S. Xu, New Res. Carbon Mater., 2014, 29, 1.
5. W. M. Yang, L. K. Liu, W. Zhou, W. Z. Xu, Z. P. Zhou and W. H. Huang, Appl. Surf. Sci., 2012, 258, 6583.
6. P. P. Xu, W. Z. Xu, X. J. Zhang and Y. S. Yan, Microchim. Acta, 2010, 171, 441.
7. F. Adam, H. Muller, A. Al-Hajji, A. Bourane and O. Koseoglu, Energy Fuels, 2015, 29, 2312.
8. M. C. Capel-Sanchez, J. M. Campos-Martin and J. L. G. Fierro, Energy Environ. Sci., 2010, 3, 328.
9. S. Nunez, J. Escobar, A. Vazquez, J. A. D. L. Reyes and M. Hernandez-Barrera, Mater. Chem. Phys., 2011, 126, 237.
10. Y. Muhammad and C. X. Li, Fuel Process. Technol., 2011, 92, 624.
11. Z. Song, T. Zhou, J. N. Zhang, H. Y. Cheng, L. F. Chen and Z. W. Qi, Chem. Eng. Sci., 2015, 129, 69.
12. N. F. Nejad and A. A. M. Beigi, Pet. Sci., 2015, 12, 330.
13. M. A. Dinamarca, C. Ibacache-Quiroga, P. Baeza, S. Galvez, M. Villarroel, P. Olivero and J. Ojeda, Bioresour. Technol., 2010, 101, 2375.
14. F. Davoodi-Dehaghani, M. Vosoughi and A. A. Ziaee, Bioresour. Technol., 2010, 101, 1102.
15. L. A. Tom, C. L. Gerard, C. M. Hutchison and A. S. Brooker, Microchim. Acta, 2012, 176, 375.
16. A. S. Ogunlaja, M. J. Coombes, N. Torto and Z. R. Tshentu, React. Funct. Polym., 2014, 81, 61.
17. W. Z. Xu, W. Zhou, P. P. Xu, J. M. Pan, X. Y. Wu and Y. S. Yan, Chem. Eng. J., 2011, 172, 191.
18. W. Z. Xu, W. Zhou, W. H. Huang, J. M. Pan, H. Li, X. Y. Wu and Y. S. Yan, Microchim. Acta, 2011, 175, 167.
19. W. M. Yang, W. Zhou, W. Z. Xu, H. Li, W. H. Huang, B. Jiang, Z. P. Zhou and Y. S. Yan, J. Chem. Eng. Data, 2012, 57, 1713.
20. W. M. Yang, L. K. Liu, Z. P. Zhou, Y. Cao and W. Z. Xu, Res. Chem. Intermed., 2015, 41, 2619.
21. T. P. Hu, Y. M. Zhang, L. H. Zheng and G. Z. Fan, J. Fuel Chem. Technol., 2010, 38, 722.
22. W. F. Liu, L. Qin, Y. Z. Yang, X. G. Liu and B. S. Xu, Mater. Chem. Phys., 2014, 148, 605.
23. W. F. Liu, X. G. Liu, Y. Z. Yang, Y. Zhang and B. S. Xu, Fuel, 2014, 117, 184.
24. M. S. Zhang, J. R. Huang, P. Yu and X. Chen, Talanta, 2010, 81, 162.
25. B. Rezaei and O. Rahmanian, J. Appl. Polym. Sci., 2012, 125, 798.
26. F. F. Duan, C. Q. Chen, G. Z. Wang, Y. Z. Yang, X. G. Liu and Y. Qin, RSC Adv., 2014, 4, 1469–1475.
27. Y. Fang, D. Gu, Y. Zou, Z. X. Wu, F. Y. Li, R. C. Che, Y. H. Deng, B. Tu and D. Y. A. Zhao, Angew. Chem., 2010, 122, 8159.
28. Y. Kim, C. Y. Cho, J. H. Kang, Y. S. Cho and J. H. Moon, Langmuir, 2012, 28, 10543.
29. X. Zhao, W. Li, S. S. Zhang, L. H. Liu and S. X. Liu, Mater. Chem. Phys., 2015, 155, 52.
30. L. Zhou, K. Dong, Z. W. Chen, J. S. Ren and X. G. Qu, Near-infrared absorbing mesoporous carbon nanoparticle as an intelligent drug carrier for dual-triggered synergistic cancer therapy, Carbon, 2015, 82, 479–488.
31. Y. L. Wang, B. B. Chang, D. X. Guan and X. P. Dong, J. Solid State Electrochem., 2015, 19, 1783.
32. Y. L. Liu, C. X. Shi, X. Y. Xu, P. C. Sun and T. H. Chen, J. Power Sources, 2015, 283, 389.
33. W. M. Yang, L. K. Liu, Z. P. Zhou, H. Liu, B. Z. Xie and W. Z. Xu, Appl. Surf. Sci., 2013, 282, 809.
34. L. Qin, W. F. Liu, Y. Z. Yang and X. G. Liu, Monatsh. Chem., 2015, 146, 449.
35. Y. Z. Yang, X. G. Liu, M. C. Guo, S. Li, W. F. Liu and B. S. Xu, Colloids Surf., A, 2011, 377, 379.
36. Q. R. Li, K. G. Yang, Y. Liang, B. Jiang, J. X. Liu, L. H. Zhang, Z. Liang and Y. K. Zhang, ACS Appl. Mater. Interfaces, 2014, 6, 21954.
37. S. Kumagai, H. Ishizawa and Y. Toida, Fuel, 2010, 89, 365.
38. M. X. Li, H. T. Wang, S. J. Wu, F. T. Li and P. D. Zhi, RSC Adv., 2012, 2, 900.
39. F. N. Memon, H. F. Ayyilidiz, H. Kara, S. Memon, A. Kenar, M. K. Leghari, M. Topkafa, S. T. H. Sherazi, N. A. Memon, F. Durmaz and I. Tarhan, Chemom. Intell. Lab. Syst., 2015, 146, 158.
40. S. Chowdhury, R. Mishra, P. Saha and P. Kushwaha, Desalination, 2011, 265, 159.
41. R. A. Ahmad and A. M. Awwad, J. Chem. Eng. Data, 2010, 55, 3170.
42. A. Adamczuk and D. Kolodynska, Chem. Eng. J., 2015, 274, 200.
43. F. B. O. Daoud, S. Kaddour and T. Sadoun, Colloids Surf., B, 2010, 75, 93.
44. C. Yu, X. M. Fan, L. M. Yu, T. J. Bandosz, Z. B. Zhao and J. S. Qiu, Energy Fuels, 2013, 27, 1499.

---