Preparation of rGO/ZrP as a new adsorbent in dibenzothiophene removal from n-decane with high capacities and good regenerability

Abstract

In this work, reduced graphene oxide (rGO)/zirconium phosphate (ZrP) nanosheets were prepared by intercalation of graphene oxide (GO) between ZrP layers followed by its reduction to rGO. Then, the adsorption behavior of the rGO/ZrP toward dibenzothiophene (DBT) was investigated. rGO/ZrP nanosheets were characterized by FT-IR, FESEM, Raman, BET and XRD analyses. The FESEM technique revealed the nano-structure of the rGO/ZrP composite and the composite has a high specific surface area of 321 m² g⁻¹. ZrP was used to prevent the reunion of reduced graphene oxide (rGO) during adsorption and storage, also increasing the specific surface area of the adsorbent. rGO/ZrP could decrease the DBT concentration from 500 ppmw to 322 ppmw after 4 h and the maximum adsorption capacity of DBT onto the rGO/ZrP was 46.6 mg g⁻¹ adsorbent. The effect of desulfurization temperature, initial concentration of DBT, duration of adsorption and regeneration of the adsorbent was investigated. The DBT removal efficiency of the rGO/ZrP adsorbent was nearly constant during the five consecutive cycles of the adsorption–desorption process. π–π interactions between rGO sheets and DBT molecules are greatly enhanced due to the pillared structure of the rGO/ZrP adsorbent. Desulfurization results were very satisfactory and rGO/ZrP could serve as a promising sorbent for the removal of DBT from petroleum.

---

1 Introduction

Deep removal of sulfur contaminants from fuel has become a very crucial and important subject for the petroleum refining industry due to increasingly stringent environmental regulations in many countries.^1–3 The most widely used technique for removing sulfur from petroleum and petroleum processing products is hydrodesulfurization (HDS),^4–6 which operates at 290–370 °C and 1.38–5.17 Mpa. HDS is highly efficient in removing (di) sulfides and mercaptans. However, it is less effective for refractory molecules such as thiophene, benzothiophene, dibenzothiophene (DBT) and their derivatives.^3,7 Moreover, ultra-deep desulfurization by HDS required substantial hydrogen consumption and inevitably leads to a significant loss in octane number due to olefin saturation.³ To avoid the mentioned problems, other desulfurization methods such as adsorption,^3,7–11 reactive adsorption,¹² biodesulfurization,^13,14 oxidation,^15–17 extraction,^18,19 and ionic liquids which are usually associated with oxidation and extraction methods^20–22 have been reported during the last few decades. Among these methods, adsorption desulfurization has attracted world spread attention for selectively adsorbing sulfur compounds from liquid fuel at mild conditions (<100 °C, ambient pressure, no hydrogen requirement); hence, this method might be the most cost effective method for the removal of organosulfur compounds.^2,23

Different sorbents used for adsorption desulfurization mainly are activated carbons,^9,10,23 (modified) zeolites,^2,7,10,24,25 metal organic frameworks (MOFs),^26–28 ionic liquids,^20–22 metal oxides,^9,29 microporous coordination polymers (MCPs),³⁰ and other innovative systems that have been recently introduced.^31–33

The key point is that most of these sorbents suffer from low capacity of sorption,^2,26 special operational conditions,^13,14 insufficient selectivity for adsorption of sulfur containing aromatic compounds compared to polyaromatics,^2,26 and most importantly, non-regeneration ability of sorbent by heating or by purging with a solvent.²⁶

In addition, there are a lot of data in literature on π-complexation and π-interaction between aromatic organosulfur compounds and sorbent due to their role in increasing the adsorption capacity. In most of these adsorption systems, presence of metal cations such as Cu(I), Ag(I) and Ni(II) is crucial to create π-complexation with aromatic organosulfur compounds.^{2,7,9,34–38} As mentioned before, the main problem of these sorbents containing metal cations is the lack of regeneration because of degradation, leaching or conversion of metal cations to useless valence forms.^7,9,28,36,38 Therefore, introduction of adsorption systems which do not need metal cations is very important from aspect of reusability of sorbent.^30,31 It is worth noting that regeneration of sorbent without using chemical solvents and by heating sorbent is very interesting environmentally and economically.

On the other hand, graphene and functionalized graphene has attracted considerable attention in different chemical and electrochemical fields due to their high surface area to volume, high thermal and mechanical stability, easy preparation, etc.^39–41 These materials were used frequently in adsorption applications such as adsorption of heavy metals,⁴² CO₂ and SO₂,⁴³ and aqueous organic pollutants.^39,44

Moreover, zirconium phosphate (ZrP) is an inorganic ion exchange material that has layered structure with tunable interlayer distance that can intercalate a wide variety of exchangers, catalysts, sorbents and carriers.^44–48 Recently, Wu et al. has reported adsorption of methylene blue by using graphene which has been intercalated between zirconium phosphate layers to prevent overlapping of graphene lamella and increasing the surface area of sorbent.⁴⁴ We used this idea by little modification to be used in desulfurization of dibenzothiophene (DBT) from n-decane (Scheme 1). We synthesized θ-ZrP (ZrP in paste form) and used it to prepare rGO/ZrP nanosheets. As a new sorbent, rGO/ZrP showed high adsorption capacity in DBT removal and good regenerability up to 5 cycles with little decrease in DBT removal capacity.

Scheme 1 Schematic representation of zirconium phosphate-pillared reduced graphene oxide as a new adsorbent in desulfurization of dibenzothiophene.

2 Materials and methods

2.1 Materials

Zirconyl chloride octahydrate (ZrOCl₂·8H₂O), n-decane, phosphoric acid (H₃PO₄, 85% v/v), concentrated H₂SO₄, hydrochloric acid (HCl, 37%), H₂O₂, KMnO₄ and graphite powder (extra pure) were purchased from Merck Co. Dibenzothiophene and hydrazine monohydrate (64–65 wt% in water) were obtained from Aldrich. All the materials were used without further purification.

2.2 Preparation of GO and ZrP

GO was synthesized from graphite powder by a modified Hummers method.⁴² Briefly, 2 g of graphite and 2 g of NaNO₃ were placed in a 500 mL flask and 92 mL of concentrated H₂SO₄ was added slowly with constant stirring, followed by the slow addition of 12 g of KMnO₄ in an ice-water bath. The solution was stirred for 3 h, and then the temperature was reached to room temperature (30 °C) and the reaction was continued for another 4 h. Then, 200 mL of water was added to the reaction mixture and the temperature of the suspension was maintained at 95 °C for 30 min. Next, 20 mL of H₂O₂ (30%) was added, so that color change from brown to pall yellow was obvious. Process continued by centrifugation at 6000 rpm for 15 min and washing the solid phase three times with 0.1 M HCl. The sample was then washed with deionized water until the pH of the solution became neutral. Then, the product was exfoliated in DI water using ultrasonication so that GO was dispersed homogenously without deposition. Finally, water evaporated under vacuum at 50 °C and dark brown GO powder was prepared.

ZrP was synthesized as previously reported by Diaz et al.⁴⁷ The typical procedure includes mixing of 300 mL of a 0.05 M ZrOCl₂·8H₂O aqueous solution by 300 mL of a 35 wt% H₃PO₄ solution. The resulting mixture was refluxed with constant stirring at 94 °C for 3 days. After completion of the reaction, product was centrifuged and washed several times with DI water resulting in a paste material. This material has named θ-ZrP which has intense peak in XRPD pattern at low angles rather than powder form of ZrP (α-ZrP) due to increasing of distance between the layers as a result of the presence of water molecules.⁴⁷

2.3 Fabrication of rGO/ZrP

The rGO–ZrP was prepared as previously reported by Wu et al.⁴⁴ with little modification as follows: 2 g θ-ZrP was dispersed in 50 mL ultrapure water to which 1 g GO, that was already well sonicated, slowly added under vigorous sonication at 60 °C for 6 h (Misonix Ultrasonic XL-2000 Series, Raleigh, NC, USA, 100 W). After that, 250 mL hydrazine monohydrate (50 wt% in water) was added and the mixture was refluxed for 48 h at room temperature. Finally, a black homogeneously dispersed mixture was prepared which filtered and repeatedly washed by DI water and dried at room temperature overnight. Then, the powder product was put into a crucible and was heated to 450 °C (heating rate of 2 °C min⁻¹) and remained 120 min at this temperature to completely burn the chemicals away. After discontinuation of thermal treatment and cooling to room temperature, the resulting product was referred to the rGO/ZrP.

2.4 Characterization techniques

FTIR spectra were recorded on a FTIR spectrophotometer (Jasco 680-plus) in the range of 400–4000 cm⁻¹ using KBr pellets. The surface morphology was observed by field emission scanning electron microscopy (FESEM) (HITACHI S-4160) after coating the samples with gold film using an acceleration voltage of 30 kV. Powder XRD analyses were performed using a diffractometer with Cu anode (Philips X'pert, PW3040, Netherland), running at 40 kV and 30 mA, scanning from 2° to 40° at 3° min⁻¹. Raman spectra were obtained on a Senterra (2009) (Bruker, Germany) confocal microscopy Raman spectrometer with a 785 nm wavelength laser excitation. Adsorption–desorption isotherms of nitrogen at 77 K were measured using a NOVAWin2, version 2.2 (Quantachrome instruments). Prior to the adsorption–desorption measurements, the sample was degassed in vacuum at 150 °C for 4 h. Thermogravimetric analysis (TGA) was conducted with a STA503 win TA (Bahr-Thermoanalyse GmbH, Hüllhorst, Germany) under N₂ flow; the temperature range of the measurements was 35–800 °C and the scanning rate was 10 °C min⁻¹.

2.5 Adsorption experiments

The adsorption tests of DBT onto the rGO/ZrP were performed in batch experiments at room temperature (30 °C). A series of 5 mL DBT in n-decane solution with different initial concentrations (100 to 1000 ppmw) in 10 mL vials were used and each flask was loaded with rGO/ZrP at various mass loadings (0.01–0.1 g). The vials were vibrated in a reciprocating vibrator at given speed. Then, at a given time intervals, the suspension was centrifuged at 5000 rpm for 10 min to separate liquid from solid phase, and then very small amount of the supernatant liquid was removed for analysis of DBT concentration by GC analyzer. The rest was returned to the initial solution and procedure repeated several times (for investigation of time effect on desulfurization).

The quantitative analysis of DBT was carried out on Agilent 6890 GC (Agilent technologies, USA) with 50 m capillary column equipped with an FID detector and high purity argon as carrier gas. The injector and detector temperatures were maintained at 280 °C and column temperature was kept at 100 °C for 1 min, then increased at a 10 °C min⁻¹ rate to 270 °C and held for 5 min at this temperature. 1 μL of the effluent solutions was injected to the GC for analysis. We used a calibration curve to achieve correct GC results.

To obtain the normalized concentration (c/c₀), detected content of DBT (c) was divided by its initial content (c₀). Then, normalized sulfur concentration versus the area under the peak in GC graph was plotted to prepare breakthrough curves. The adsorption capacity was obtained by integral calculus.²⁹ All experiments were done in triplicate and the results were averaged.

The amount of DBT adsorbed onto the rGO/ZrP sorbent was calculated from the equation:

where q_c is the amount of the adsorbed DBT, V is the volume of the liquid phase, C₀ is the initial concentration of DBT, C_e is the concentration of DBT at equilibrium, and m is the amount of adsorbent.³⁴

For regeneration tests, the rGO/ZrP after adsorption process heated at 450 °C for 2 h to completely burn the chemicals away. Then, adsorption–desorption experiments were repeated four other times after heat treatments and cooling down the adsorbent to room temperature. The adsorption capacity was the difference between the initial and final concentration of DBT solution, and the desorption capacity was the difference between the initial and final weights of the rGO/ZrP before and after adsorption.

The Langmuir and Freundlich adsorption isotherm models were used to investigate the adsorption data. The linearized form of the Langmuir model assumes monolayer adsorption of solute onto the adsorbent. In this model, adsorption only occurs at a certain number of sorbent's sites and there is no lateral interaction or steric hindrance between the adjacent or non-adjacent adsorbed molecules. Linearized form of this model can be formulated as follows:

where C_e is equilibrium concentration of the solute (mg L⁻¹) and q_e is the solid phase equilibrium concentration (mg solute per g adsorbent). Q_max is the maximum sorption capacity per unit weight of the adsorbent (mg g⁻¹), and K_L is a constant related to binding energy of the sorption system (L mg⁻¹). The K_L and Q_max can be obtained from the intercept and slop of the linear plots when C_e/q_e obtains in a fairly straight line.⁴⁹

The linear form of Freundlich isotherm model which unlike the Langmuir isotherm is applied to multilayer adsorption, with non-uniform distribution of adsorption heat and affinities over the heterogeneous surface can be represented by logarithmic eqn (3):

where n and K_F are the Freundlich isotherm constants which are corresponded to the adsorption capacity and intensity of a given adsorbent, respectively.^44,49

3 Results and discussion

3.1 FTIR spectroscopy analysis

FTIR spectra of ZrP, GO, rGO and rGO/ZrP nanosheets are shown in Fig. 1. In the spectrum of ZrP, the vibrational frequencies at 3596, 3512, and 3166 cm⁻¹ are related to the O–H stretching of water molecules in the interlayer spaces of ZrP,⁴⁶ and P–OH groups in ZrP, respectively; the band at 1624 cm⁻¹ is attributed to the bending vibrations of O–H groups. The peaks at 1252 and 975 cm⁻¹ are associated with the in-plane and out-of-plane vibration of the P–OH groups,⁴⁶ whereas the bands centered at 1070 are related to the stretching of P–O in the PO₂ groups.^44,46 The two peaks at 598 and 526 cm⁻¹ are ascribed to the vibration of Zr–O.⁴⁶

Fig. 1 FT-IR spectra of ZrP, GO, rGO, and rGO/ZrP.

FTIR spectrum of GO shows a broad and intensive peak at 3420 cm⁻¹ that is attributed to O–H stretching vibration of –OH, and carboxyl groups of GO and water molecules entrapped inside GO. The C=O, –OH, C=C and C–O–C vibration frequencies of GO after oxidation of graphite were detected at 1727, 1626, 1384 and 1048 cm⁻¹, respectively.⁴⁴ As can be seen, similar to the pristine graphite, after reduction of GO to rGO, peaks around 3420, 1048 and 1626 cm⁻¹ remained nearly unchanged and other peaks were disappeared or became very weak.⁴⁴ FTIR spectrum of rGO/ZrP indicated both strong peaks of rGO and ZrP but some weak peaks became weak, disappeared or coated by another stronger peak such as peaks at 3166 and 975 cm⁻¹ of ZrP. Moreover, by intercalation of GO between ZrP layers and reduction to rGO, the peaks of the intercalated water molecules in the region of 3594 and 3512 cm⁻¹ disappeared.⁴⁶ We could mention certainly with these observations that rGO has been intercalated and accompanied to large extent with ZrP.

3.2 XRD and Raman spectroscopy analyses

Fig. 2a indicates the XRPD patterns of ZrP, rGO, rGO/ZrP and physical mixture of rGO–ZrP (phys. mix. of rGO–ZrP). The interlayer spacing of pristine α-ZrP is 7.6 Å, which corresponds to the distance between two adjacent layers of ZrP; characteristic peaks of α-ZrP are located at 2θ = 11.6°, 19.6°, 24.7° and 33.7°. The peak at 11.6° is related to a d₀₀₂ basal spacing of 0.76 nm.^44,47 The broad peak at 2θ ≈ 25° in the XRPD pattern of rGO, represented the interlayer spacing of 0.34 nm. This value was slightly larger than that of graphite due to the residual functional groups (–OH, –COOH) that exist between the rGO layers (Scheme 1).⁴⁴

Fig. 2 (a) XRD patterns of ZrP, rGO, rGO/ZrP and phys. mix. of rGO–ZrP; (b) Raman shift of ZrP, phys. mix. of rGO–ZrP and rGO/ZrP.

Insertion of rGO sheets between the layers of ZrP should increase interlayer spacing of pristine α-ZrP from 7.6 Å up to 9.5 Å, that is observed clearly in the XRPD pattern of the rGO/ZrP and Scheme 1. For rGO/ZrP, the peaks at 2θ = 9.2°, 9.9° and 10.8° are corresponding to the d₀₀₂ basal spacing of 0.95 nm, 0.89 nm and 0.82 nm, respectively, due to the mixed-phase intercalation of rGO into ZrP.⁴⁷ It is worth noting that by intercalation of rGO into ZrP, the broad peak of rGO in 2θ ≈ 25° could also be seen in rGO/ZrP nanosheets. To better demonstrate the proper accomplishment of intercalation process, we mixed physically rGO and ZrP (rGO : ZrP = 1 : 2, similar to the rGO/ZrP) in a mortar and pestle and acquired XRD pattern of this mixture, too. As can be seen, there is no significant change in XRD pattern of phys. mix. of rGO–ZrP compared with ZrP, unless very little broadening of peak in 2θ ≈ 25° that is attributed to rGO which is observable in rGO/ZrP pattern, too. The absence of new peak in domain below 11.6° or shift of this peak to lower domains means that intercalation process couldn't be done not at all in phys. mix. of rGO–ZrP. This has two major reasons: (1) powder ZrP has lower interlayer distance rather than paste ZrP or suspension form of ZrP in water;⁴⁷ (2) the layers of rGO in powder form, reunion over time.⁴⁴ Due to the aforementioned reasons, intercalation of powder rGO between powder ZrP layers by physical mixing seems impossible or highly unlikely that was demonstrated more by no significant change in adsorption capacity of phys. mix. of rGO–ZrP in comparison with rGO (Section 3.5.1).

It is worth noting that we used hydrated phase of α-ZrP, i.e., θ-ZrP instead of α-ZrP. By doing so, there was no need to separate and dry up synthesized ZrP, which will save time and costs. It is important to note that, θ-ZrP has interlayer distance more than α-ZrP and can easily intercalate large molecules without any preintercalators (data not shown).⁴⁸

Raman spectra of rGO, rGO/ZrP and phys. mix. of rGO–ZrP are shown in Fig. 2b. In the spectrum of rGO two characteristic peaks centered at 1310 cm⁻¹ (D band) and 1588 cm⁻¹ (G band) are related to the defects in the graphitic sp² carbon structures and the phonon vibrations of sp² bonded carbon atoms in a two-dimensional hexagonal lattice of rGO, respectively.⁵⁰ In addition, the ratio of the two peak intensities (I_D/I_G) is proportional to the level of disorder and size of graphitic domains.⁵¹ As can be seen in Fig. 2b, for rGO, phys. mix. of rGO–ZrP and rGO/ZrP, I_D/I_G are 1.48, 1.52, and 1.65, respectively. The higher I_D/I_G value for rGO/ZrP means the formation of smaller sp² graphitic domains on reduction of GO and higher reduction degree of the GO due to better and much separation of GO layers by ZrP, which in turn results in hindering the self-reunion of the graphene sheets⁵¹ that was more confirmed by adsorption data (Section 3.5.1).

3.3 FESEM analysis

The morphology of the rGO/ZrP nanosheets is shown in Fig. 3. As can be seen, the average particle diameter is lower than 100 nm and rGO/ZrP exhibited a highly rough surface with densely packed structure (Fig. 3a). Layered and porous nature of nanosheets is clearly observed in Fig. 3b. Highly rough surface with nano-dimensions and creating pores with approximate size of 2–10 nm (confirmed by using BET data, Section 3.4) would provide more active sites for adsorptive desulfurization and help to enhance the adsorption activity of rGO/ZrP by increase of surface to volume ratio.⁴⁴ Therefore, one would expect to observe an increasing in adsorption capacity of rGO/ZrP nanosheets rather than free rGO, however, intercalation process has also important role in this interesting improvement in desulfurization capacity (Section 3.5.1).

Fig. 3 FESEM images of rGO/ZrP (a) nano-structure nature, and (b) layered and porous nature of the nanosheets.

3.4 N₂ adsorption–desorption isotherm

The N₂ adsorption–desorption isotherm and the pore size distribution of the rGO/ZrP nanosheets from BJH method adsorption are presented in Fig. 4a and b, respectively. The specific surface area of the rGO/ZrP nanosheets was 321 m² g⁻¹, and showed an adsorption–desorption isotherm similar to the type IV which is considered to be indicative of adsorption in mesoporous solids. The H2 hysteresis loop of rGO/ZrP nanosheets is characteristic of solids consisting of particles crossed by nearly cylindrical channels or made by aggregates or agglomerates of spherical particles, that is more confirmed by FESEM images (Fig. 3a and b). Non-uniform size or shape of pores and different sizes and dimensions of pore mouth and pore body are characteristics of this hysteresis that is demonstrated more by extension of hysteresis loop in a wide range of relative pressure from 0.1 to 0.9.⁵²

Fig. 4 (a) Nitrogen adsorption–desorption isotherms (77 K) of rGO/ZrP; (b) pore size distributions of rGO/ZrP.

Theoretical value of surface area of rGO/ZrP is 2620 m² g⁻¹ that is substantially more than practical value (321 m² g⁻¹), the reason is incomplete exfoliation and aggregation during reduction process because of the unavoidable van der Waals force and to π–π bonding interactions between rGO sheets. This interactions are very stronger in free rGO sheets that have not disincentive factor such as ZrP to interrupt reunion and re-overlap their sheets over time, which would cause less BET specific surface area than rGO/ZrP (data not shown), as a result, rGO has indicated less desulfurization capacity than rGO/ZrP nanocomposite (Section 3.5.1).⁴⁴

3.5 Adsorption studies of DBT on rGO/ZrP

3.5.1 Investigation of different initial concentration of DBT on adsorption capacity. Fig. 5 indicates the effect of different initial concentrations of DBT on maximum adsorption capacities of rGO/ZrP and rGO adsorbents. As could be seen, the maximum adsorption capacities of DBT onto the rGO, rGO/ZrP and phys. mix. of rGO–ZrP (rGO : ZrP = 1 : 2) were 20.4 mg g⁻¹, 46.6 mg g⁻¹ and 21.6 at 30 °C, respectively. These results indicated that the adsorption quantity of DBT onto the rGO/ZrP was dependent on the initial concentration of DBT. The optimal concentration of DBT in n-decane for desulfurization by using 0.05 g of rGO, rGO/ZrP or phys. mix. of rGO–ZrP as sorbents was 800 ppmw, but due to little difference between 500 ppmw and 800 ppmw, we used 500 ppmw of DBT in other investigations. The main reason for lower (about half) adsorption capacity of rGO in comparison with rGO/ZrP might be related to the reunion and re-overlap of rGO sheets, while in the composite of rGO/ZrP this process could be avoided by ZrP.⁴⁴ As can be seen, there is no much difference between desulfurization capacity of phys. mix. of rGO–ZrP and rGO, because intercalation process was not accomplished any way by simple abrasion of these two matter as proved by XRD pattern of this mixture (Section 3.2). The improvement in adsorption performance of the phys. mix. of rGO–ZrP is likely due to crush, shrink and better distribution of rGO particles in DBT solution because of abrasion with ZrP.

Fig. 5 Effects of the initial concentration on adsorption of DBT onto the rGO, rGO/ZrP and phys. mix. of rGO–ZrP after 24 h.

3.5.2 Effect of adsorption time and temperature on adsorption of DBT onto the rGO/ZrP. The effects of adsorption time and temperature on adsorption of DBT onto the rGO/ZrP were shown in Fig. 6. It could be seen that concentration of DBT was decreased sharply at the first 30 min, then reduced slowly until 4 h that the equilibrium was reached. The effect of temperature on DBT adsorption was shown in Fig. 6 (inset). After 4 h, the concentration of DBT solution from initial value of 500 mg L⁻¹ has reached to 281, 276 and 273, respectively, at the temperatures of 30, 40 and 50 °C. It can be concluded that higher temperature facilitated the adsorption of DBT onto the rGO/ZrP. The increase of temperature might increase energy of DBT molecules to penetrate into more internal active sites of rGO/ZrP nanosheets and result in higher adsorption capacity of the adsorbent.

Fig. 6 The effects of adsorption time and temperature (inset) on adsorption of DBT onto the rGO/ZrP.

3.5.3 Adsorption–desorption property and regeneration capability of the rGO/ZrP. We used TGA test to better indicate the adsorption–desorption capability of our adsorbent (Fig. 7a). As can be seen, TGA curve of rGO/ZrP shows two major weight losses below 800 °C: first, in 50–200 °C and the second one in the range of 400–600 °C that correspond to the evaporation of free (absorbed) interlayer water and condensation (dehydroxylation) of ZrP and rGO lattice, respectively.^44,46

Fig. 7 (a) TGA curves of rGO/ZrP and DBT/rGO/ZrP; (b) adsorbed and desorbed quantities (mg DBT per g rGO/ZrP) for 5 cycles using rGO/ZrP as adsorbent.

By contrast, the TGA curve of DBT/rGO/ZrP shows three major weight losses below 800 °C: first, evaporation of absorbed water and gases below 200 °C; second decomposition of the DBT between 200 and 500 °C; and third, condensation (dehydroxylation) of the rGO and ZrP lattices between 500 and 700 °C. By comparing the total weight losses of rGO/ZrP and DBT/rGO/ZrP, the amount of additional weight loss in DBT/rGO/ZrP was about 11% which corresponds to the adsorbed DBT onto the rGO/ZrP adsorbent. This value is in good agreement with the results of adsorption capacity (Fig. 5).

To regenerate the saturated rGO/ZrP adsorbent, it was simply put into a crucible and heated at 450 °C for 120 min to completely burn the chemicals away. Then, it was used to investigate adsorption capacity and this procedure was repeated four other times. The adsorption capacity was the difference between the initial and final concentrations of DBT solution, while the desorption capacity was the difference between the initial and final weights of the pure and saturated rGO/ZrP.

Regeneration results were shown excellent regenerability of the rGO/ZrP adsorbent after 5 cycles (Fig. 7b). In this regard, the average adsorption capacity of the rGO/ZrP adsorbent was changed from 46.6 mg DBT per g adsorbent to 43.8 after fifth regeneration cycle, while corresponding desorption capacities also had little reduction from 41.1 to 37.3 mg DBT per g adsorbent from first to fifth desorption cycle.

To better understand the importance of the introduced adsorbent, we compared the results obtained in this work with some of the previously reported data (Table 1). The best conditions of the reported adsorbents were considered. The adsorption capacities were entered as mentioned in the articles or calculated with good approximation.

Table 1 Comparison of desulfurization capacities of rGO/ZrP adsorbent with some of the previously reported adsorbents

Adsorbent	Adsorbate	Conditions	Adsorption capacity (mg g^−1)	Regenerability investigation	Ref.
a Metal organic framework.b Poly-methylbenzene.c Dimethyldibenzothiophene.d Porous vanadium benzene dicarboxylate.
Cu(I)–zeolite Y	DBT	30 °C	0.19	Yes	2
Cu–zeolite Y	Thiophene	90 °C, 0.1 atm	0.32	No	7
Activated carbon (CAA)	Sulfur containing feedstock	30 °C	1.7	No	23
Cu^+-MOF^a	Benzothiophene	25 °C	75	No	28
Ag/PMB^b-3	DBT	25 °C	11.8	No	32
Nanoporous activated carbon	4,6-DMDBT^c	25 °C	20	No	34
Ni–Cu/γ-Al2O3	DBT	200 °C	4	Yes	35
CuCl2/pVBDC^d (MIL-47)	Benzothiophene	25 °C	310	No	36
Ag/TiO2–Al2O3	Benzothiophene	Room temp.	13	No	37
Cu(I)–zeolite Y	Sulfur containing diesel fuel	Room temp.	0.122	No	38
rGO/ZrP	DBT	30 °C	46.6	Yes	Present work

---

As can be seen, most of the reported adsorbents have adsorption capacities lower than the rGO/ZrP. However, cases which are more efficient than rGO/ZrP, did not investigate the regenerability of the adsorbent more likely due to the decomposition of the adsorbent, especially change of state of metal ions over the adsorbent.^28,36 Moreover, the regeneration cost of adsorbent in some works is high and the procedure is complex and difficult.^2,35

3.5.4 Adsorption kinetic. Adsorption isotherms of Langmuir and Freundlich models present in Fig. 8. As can be seen, Langmuir isotherm that has R²: 0.974 (Fig. 8a), is the best model to demonstrate the adsorption of DBT onto the rGO/ZrP adsorbent, while the Freundlich isotherm has R²: 0.904 (Fig. 8b). Therefore, monolayer adsorption kinetic of DBT onto the rGO/ZrP adsorbent is proposed which occurred by strong π–π stacking interactions between the aromatic rings of DBT and the pillared rGO/ZrP surface. ZrP will prevent reunion and re-overlap of rGO sheets resulting in high adsorption capacity of the adsorbent.⁴⁴

Fig. 8 Sorption isotherms of DBT onto the rGO/ZrP adsorbent at T = 30 °C, (a) Langmuir plot; (b) Freundlich plot.

4 Conclusions

rGO/ZrP nanosheets were prepared by using simple and cost-effective procedure. FESEM technique proved nanostructure of these nanosheets. The most important advantage of intercalation of rGO between ZrP layers was to prevent reunion of rGO sheets, and as a result no change of surface area of the adsorbent over time. rGO/ZrP adsorbent could adsorb dibenzothiophene more than 45 mg g⁻¹ after 4 h. In contrast to the most previously reported adsorbents, the introduced adsorbent in this work can simply be regenerated several times without any damage to its structural framework and without significant drop in its adsorption efficiency. Moreover, it can be regenerated by heating in a furnace in air at atmospheric pressure. In situ synthesis of rGO/ZrP (synthesis of ZrP in GO suspension followed by reduction of GO to rGO) might be a desire methodology to eliminate costly and overwhelming ultrasonication operation.

Acknowledgements

Thanks are due to the Research Council of Isfahan University of Technology and Center of Excellency in the Chemistry Department of Isfahan University of Technology for supporting of this work.

References

1. C. Song, Catal. Today, 2003, 86, 211–263.
2. W. Li, J. Xing, X. Xiong, J. Huang and H. Liu, Ind. Eng. Chem. Res., 2006, 45, 2845–2849.
3. F. Tian, Q. Shen, Z. Fu, Y. Wu and C. Jia, Fuel Process. Technol., 2014, 128, 176–182.
4. R. Shafi and G. J. Hutchings, Catal. Today, 2000, 59, 423–442.
5. H. Shang, W. Du, Z. Liu and H. Zhang, J. Ind. Eng. Chem., 2013, 19, 1061–1068.
6. R. Singh, D. Kunzru and S. Sivakumar, Appl. Catal., B, 2016, 185, 163–173.
7. A. Takahashi, F. H. Yang and R. T. Yang, Ind. Eng. Chem. Res., 2002, 41, 2487–2496.
8. M. Muzic, K. S. Bionda, Z. Gomzi, S. Podolski and S. Telen, Chem. Eng. Res. Des., 2010, 88, 487–495.
9. D. Yi, H. Huang, X. Meng and L. Shi, Ind. Eng. Chem. Res., 2013, 52, 6112–6118.
10. P. Baltzopoulou, K. X. Kallis, G. Karagiannakis and A. G. Konstandopoulos, Energy Fuels, 2015, 29, 5640–5648.
11. A. B. S. H. Salem, Ind. Eng. Chem. Res., 1994, 33, 336–340.
12. G. Wang, Y. Wen, J. Fan, C. Xu and J. Gao, Ind. Eng. Chem. Res., 2011, 50, 12449–12459.
13. A. K. Chauhan, A. Ahmad, S. P. Singh and A. Kumar, Int. Biodeterior. Biodegrad., 2015, 104, 105–111.
14. P. Derikvand, Z. Etemadifar and D. Biria, Int. Biodeterior. Biodegrad., 2014, 86, 343–348.
15. P. de Filippis and M. Scarsella, Energy Fuels, 2003, 17, 1452–1455.
16. D. Zheng, W. Zhu, S. Xun, M. Zhou, M. Zhang, W. Jiang, Y. Qin and H. Li, Fuel, 2015, 159, 446–453.
17. A. D. Bokare and W. Choi, J. Hazard. Mater., 2016, 304, 313–319.
18. D. Julião, A. C. Gomes, M. Pillinger, L. Cunha-Silva, B. de Castro, I. S. Gonçalves and S. S. Balula, Fuel Process. Technol., 2015, 131, 78–86.
19. S. O. Ribeiro, D. Julião, L. Cunha-Silva, V. F. Domingues, R. Valença, J. C. Ribeiro, B. de Castro and S. S. Balula, Fuel, 2016, 166, 268–275.
20. M. Francisco, A. Arce and A. Soto, Fluid Phase Equilib., 2010, 294, 39–48.
21. S. A. Dharaskar, K. L. Wasewar, M. N. Varma, D. Z. Shende and C. K. Yoo, Procedia Eng., 2013, 51, 314–317.
22. A. Seeberger and A. Jess, Green Chem., 2010, 12, 602–608.
23. C. M. Rosas, L. F. R. Verduzco, F. R. M. Guevara, G. H. Tapia and L. M. R. Otal, Ind. Eng. Chem. Res., 2010, 49, 4372–4376.
24. J. J. Liao, W. R. Bao and L. P. Chang, Fuel Process. Technol., 2015, 140, 104–112.
25. H. Song, X. H. Cui, H. L. Song, H. J. Gao and F. Li, Ind. Eng. Chem. Res., 2014, 53, 14552–14557.
26. D. Peralta, G. Chaplais, A. S. Masseron, K. Barthelet and G. D. Pirngruber, Energy Fuels, 2012, 26, 4953–4960.
27. I. Ahmed and S. H. Jhung, J. Hazard. Mater., 2016, 301, 259–276.
28. N. A. Khan and S. H. Jhung, J. Hazard. Mater., 2012, 237–238, 180–185.
29. J. Kou, C. Lu, W. Sun, L. Zhang and Z. Xu, ACS Sustainable Chem. Eng., 2015, 3, 3053–3061.
30. K. A. Cychosz, A. G. W. Foy and A. J. Matzger, J. Am. Chem. Soc., 2009, 131, 14538–14543.
31. A. F. Réguillon, M. Sévignon, M. Rocault, E. Schulz and M. Lemaire, Ind. Eng. Chem. Res., 2008, 47, 9617–9622.
32. Y. Xia, Y. Li, Y. Gu, T. Jin, Q. Yang, J. Hua, H. Liu and H. Wang, Fuel, 2016, 170, 100–106.
33. I. Shimoyama and Y. Baba, Carbon, 2016, 98, 115–125.
34. X. Ma, M. Sprague and C. Song, Ind. Eng. Chem. Res., 2005, 44, 5768–5775.
35. A. Mansouri, A. A. Khodadadi and Y. Mortazavi, J. Hazard. Mater., 2014, 271, 120–130.
36. N. A. Khan and S. H. Jhung, Angew. Chem., Int. Ed., 2012, 51, 1198–1201.
37. A. H. M. S. Hussain and B. J. Tatarchuk, Fuel Process. Technol., 2014, 126, 233–242.
38. A. J. H. Maldonado and R. T. Yang, J. Am. Chem. Soc., 2004, 126, 992–993.
39. Y. Yao, Y. Cai, F. Lu, F. Wei, X. Wang and S. Wang, J. Hazard. Mater., 2014, 270, 61–70.
40. T. Kuila, S. Bose, A. K. Mishra, P. Khanra, N. H. Kim and J. H. Lee, Prog. Mater. Sci., 2012, 57, 1061–1105.
41. M. Pumera, Electrochem. Commun., 2013, 36, 14–18.
42. N. Saeidi, M. Parvini and Z. Niavarani, J. Environ. Chem. Eng., 2015, 3, 2697–2706.
43. S. Yun, H. Lee, W. E. Lee and H. S. Park, Fuel, 2016, 174, 36–42.
44. Z. Wu, L. Zhang, Q. Guan, P. Ning and D. Ye, Chem. Eng. J., 2014, 258, 77–84.
45. D. Li, C. Miao, X. Wang, X. Yu, J. Yu and R. Xu, Chem. Commun., 2013, 49, 9549–9551.
46. Y. Zhou, R. Huang, F. Ding, A. D. Brittain, J. Liu, M. Zhang, M. Xiao, Y. Meng and L. Sun, ACS Appl. Mater. Interfaces, 2014, 6, 7417–7425.
47. A. Diaz, A. David, R. Perez, M. L. Gonzalez, A. Baez, S. E. Wark, P. Zhang, A. Clearfield and J. L. Colon, Biomacromolecules, 2010, 11, 2465–2470.
48. A. Diaz, M. L. Gonzalez, R. J. Perez, A. David, A. Mukherjee, A. Baez, A. Clearfield and J. L. Colon, Nanoscale, 2013, 5, 11456–11463.
49. K. Y. Foo and B. H. Hameed, Chem. Eng. J., 2010, 156, 2–10.
50. Z. L. Cheng, Y. Y. Liu and B. C. Cao, Mater. Lett., 2016, 175, 215–218.
51. S. P. Lonkar, J. M. Raquez and P. Dubois, Nano-Micro Lett., 2015, 7, 332–340.
52. G. G. Leofanti, M. Padovan, G. Tozzola and B. Venturelli, Catal. Today, 1998, 41, 207–219.

---

Footnote

† Final version of the manuscript was written through equal contributions of all authors. The authors have given approval to the manuscript.

---