Desulfurization of liquid fuels using aluminum modified mesoporous adsorbent: towards experimental and kinetic investigations

In this study, a modified mesoporous adsorbent (MSU-S) impregnated by aluminum was used to remove the aromatic sulfur compounds from n-decane as the model fuel. Physical and chemical properties of as-synthesized adsorbent were investigated by XRD (X-Ray Diffraction), SEM (Scanning Electron Microscopy), FTIR (Fourier Transform Infrared spectroscopy) and BET (Brunauer–Emmett–Teller) method. Adsorptive desulfurization of model fuel was studied through batch and continues processes under mild temperature and normal atmospheric pressure. The equilibrium adsorption was modeled by Langmuir, Temkin, and Freundlich and the kinetics of adsorption was studied through first, second and intraparticle diffusion models. It was figured out that Temkin and the pseudo-second-order model were best fitting the adsorption equilibrium and describing the kinetics, respectively.

Impregnation. Al (NO 3 ) 3 used as an aluminum source to be loaded on MSU-S. In order to make a dispersed phase of Al 2 O 3 on the MSU-S, the required amount of metal precursor was mixed in DW. A specific amount of MSU-S was introduced to the mixture while it was mixed and left over the night at 70 °C to lose moisture. The prepared gel was desiccated at 90 °C for 12 h. After that, for the calcination process, the final product was put in the furnace and heated with the rate of 1 °C/min to reach 550 °C and preserved at this temperature for 4 h. In order to activate the synthesized material and before being used in the adsorption test, it was placed at hydrogen atmosphere at 190 °C.
Model fuel. To prepare synthesized fuel like a real fuel, n-decane as the paraffinic compound and dibenzothiophene as the sulfur source were used.
Characterization of samples. The XRD (X-ray diffraction) was studied applying wave length = 1.5418 Å, 40 kV and 40 mA of Cu Kα radiation in the domain from 1° to 10° and 20° to 80°. The Fourier transform infrared (FTIR) pattern of the adsorbent was reported on the Nicolet Impact 400. 128 scans were picked at a 4 cm −1 resolution in the mid-IR region using KBr. The Brunauer-Emmett-Teller (BET) was applied to derive surface area and average pore diameter. The textural properties of the materials were investigated by N 2 adsorption/ desorption at − 196 °C with the Belsorp-mini-instrument. SEM (scanning electron microscopy) was done on the Zeiss instrument for the morphological study of the sorbent. The chemical analysis of Energy dispersive X-ray spectroscopy (EDS) was performed with Oxford Instrument.
General adsorption tests. Batch process. Batch experiment is done to discuss equilibrium and kinetics of adsorption of DBT on Al 2 O 3 -MSU-S and then to investigate adsorption capacity, equilibrium and kinetics models. For all experiments, 0.2 g adsorbent and 0.5 g model fuel were used. The tests were performed in the capped vials and mixed with constant stirring speed of 250 rpm at room temperature and atmospheric conditions.
Continuous adsorption experimental procedure. In order to get the breakthrough and saturation amount for the adsorbent, the experiments were performed in a continuous process. For the continuous tests, DBT content in the fuel was considered to be 400 ppmw. The breakthrough /adsorption experiments were done in the reactor of fixed-bed stainless-steel (150 mm long, 10 mm internal diameter) with a layer of quartz sand. Approximately, 0.5 g of Al 2 O 3 -MSU-S was placed into the reactor. In order to obtain the saturation point the treated fuel was sampled every 5-10 min.
Sulfur concentration analysis. The sulfur amount in the synthesized fuel was defined by GC Gas Chromatograph YL6100 model with the capillary column (5 TRB, Length: 30 m, Internal Diameter: 0.32 mm, DF: 1 μm; Teknokroma.), and the FID containing He as carrier gas.

Results and discussion
Adsorbent features. XRD. Figure 1a shows the XRD low-angle spectra of the adsorbents. As it illustrate, the influence of loading aluminum decreases maximum intensity of MSU-S clearly, which depicts that crystalline phase converts to amorphous phase relatively [24][25][26] . It is obvious that impregnating of aluminum has a negative impact on mesoporous pattern of MSU-S. As illustrated in Fig. 1b, there are four maximums in the highangle area of XRD spectra of Al 2 O 3 -MSU-S, associated with the head for MSU-S hexagonal mesostructure. All these maximum points indicate the formation of crystalline phase of aluminum oxide on the adsorbent (MSU-S) after being injected on the surface 6 (Fig. 2 Fig. 3. It is obvious that there is a hysteresis loop during adsorption-desorption. According to the IUPAC classification, isotherm type IV can be used to describe hysteresis behavior of Al 2 O 3 -MSU-S that is the characteristic feature of mesoporous solid. A low-pressure hysteresis which is gained in Fig. 3 extending to the lowest achievable pressures. A step occurs in the adsorption curve between partial pressures P/Po of 0.8 to 1. Figure 4 illustrates the pore size dispensation of Al 2 O 3 -MSU-S. The sharp peak around 21 Å in this figure might be according to the internal cavity and represents the narrow pore size [28][29][30][31][32][33] .  Adsorption mechanism. As mentioned in other studies and research, sulfur compounds can be removed by certain sorbents either through the S-M bonds formation or the π-complexation. Direct sulfur-sorbent interactions by ion exchange between metal oxide (Al 3+ ) and sulfur compounds can be attributed to hard and soft acids and bases (HSAB) principle. According to this theory, hard acids prefer to connect with hard bases, and soft acids prefer soft bases. The large differences in electronegativity between hard acids and hard bases leads to strong ionic interactions. The electronegativities of soft acids and soft bases are almost the same and therefore, they have less ionic interactions and they almost make covalent connections. The interactions between hard acid-soft base or soft acid-hard base are mostly polar covalent and tend to be more reactive or less stable. The polar covalent compounds readily make either ionic or covalent compounds if they are permitted to react. Al 3+ is a hard acid and prefers to combine with DBT by the direct S-M (metal oxide) σ-bond. In this case, the strength of the direct S-M σ-bond basically depends on the charge number and ionic radius of the metal ion 21 .  www.nature.com/scientificreports/ C e (mg kg −1 ) is the content of sulfur-component material at equilibrium; K L (kg mg −1 ) and q m (mg g −1 ) are the Langmuir isotherm parameters relevant to the adsorption energy and the largest load, respectively; K F (mg 1−(1/n) / g −1 kg 1/n ) and 1/n show the Freundlich parameters relevant to the adsorption volume and also in Temkin ; R is universal constant, T is temperature, A T (L/g), and b T are Temkin constants, respectively; and q e (mg g −1 ) is the content of sulfur-component adsorbed at the equilibrium/mass of adsorbent. Equations (4)-(6) show the linearized form of the Langmuir, Freundlich and Temkin models.
K L and q m for Eq. (4), also K F and 1/n in Eq. (5), RT b T and A T for Eq. (6) can be found using slope and intercept of the plotted experimental data. Figures 6, 7, 8 illustrate the linear form of models fitted on the experimental data 2,19 . Tables 2 and 3 demonstrates the calculated parameters from regression of data and with the correspondence coefficient (R 2 ) amounts of the models for adsorption isotherms for MSU-S and Al 2 O 3 -MSU-S. It is obvious that Temkin model can be used to define the adsorption of DBT on the sorbent due to higher values of R 2 compared with other models. Based on this model, there is an indirect adsorption interaction between DBT and Al 2 O 3 -MSU-S as adsorbent. It means the heat of adsorption decreases with increasing coverage in a linear relation. Therefore, a uniform distribution of binding energies was observed in the adsorption process. Fig. 6 it is clear that by increasing the contact time between model fuel and adsorbent, the residual concentration of sulfur content decreased. This figure points out that the major and effective adsorption takes place in the first 40 min and after a while reaches to an asymptotic amount.

Kinetics of DBT adsorption. From
Pseudo-first order, pseudo-second order, and intraparticle diffusion kinetic relations have been utilized to assess the kinetics of sulfur adsorption onto Al 2 O 3 -MSU-S.
The pseudo-first order relation is shown as: q t = adsorption loads (mg/g) at the time of t, q e = adsorption loads (mg/g) at equilibrium, k 1ad = constant rate of pseudo-first order adsorption (per minute). Integrating from Eq. (7) and employing boundary limits from t = 0 to t = t and q t = 0 to q t = q t , leads to the following equation:  www.nature.com/scientificreports/ As shown in Eq. (8) if ln (q e − q t ) vs. time is plotted, k 1ad and lnq e could be defined through slope and intercept of the model, respectively. Equation (9) shows the pseudo-second-order kinetic reltion 34-36 : where k 2ad expresses the rate constant (gr/mg.min). After being integrated and employing boundary values (t = 0 to t = t and q t = 0 to q t = q t ), equation can be derived in the following 35-38 : As shown in Eq. (10) if t/q t vs. t is plotted, k 2ad and q e are defined through slope and intercept of the model. The intraparticle pattern which was presented by Weber and Morris is indicated as follows 39,40 : where k i shows the rate constant (mg/g min 0.5 ) and C is a constant (mg/g). As shown in Eq. (11) if q t vs. t 0.5 is plotted, k i and C can be specified through slope and intercept of the model, respectively. Tables 4 and 5 illustrates the constant parameters and R 2 values of models which are calculated by curve fitting of experimental data for MSU-S and Al 2 O 3 -MSU-S. As it is observed clearly, the best fitting for adsorption experimental data is the pseudo-second order pattern with maximum correlation coefficients in comparison with the intraparticle diffusion pattern and pseudo-first order models 18,39,40 . Figure 7 shows the performance of pseudo second order model to predict experimental data. Regarding the experimental and theoretical study, k 2 , the pseudo-second order constant is related to the primary concentration of the adsorbate. From this model, it can be suggested that the adsorption procedure is in two-steps. The first stage exhibits the film diffusion from mixture to the outer surface of adsorbent and the second step indicates diffusion into the rough surface and abundant voids. One of the major advantages of this model is capability of modeling adsorption with significant change in adsorbate concentration which makes it reliable. Figure 8 shows the breakthrough graphs of adsorption of DBT on Al 2 O 3 -MSU-S at ambient temperature, atmospheric pressure and (8) ln q e − q t = lnq e − k lad t (9) dq t dt = k 2ad q e − q t 2 (10)  www.nature.com/scientificreports/ 1.0186 per hour LHSV. For these experiments 2.92 g of the model fuel/gram of sorbent (g-F/g-A) was used in the fixed-bed adsorber. When saturation point is reached, the C/C o value for DBT rises strongly. In order to find saturation capacity of Al 2 O 3 -MSU-S, integral calculus of the breakthrough curve is used. It was observed that adsorption load of presented sorbent to adsorb DBT from model fuel is 0.072 mmol/g 7 . Table 6 shows the adsorption capacity of different loads of metals on MSU-S adsorbent. It is obvious that the maximum adsorption load of DBT is related to Al 2 O 3 -MSU-S adsorbent.

Conclusion
In this work, MSU-S was synthesized, modified with aluminum and characterized to be applied in adsorptive desulfurization of model fuel. Modification of present MSU-S by the aluminum had negative effect which decreased the surface area for adsorption. Loading aluminum ions on the MSU-S enhanced adsorption selectivity through ion-exchange in the adsorbent. Comparing between aluminum modification and other metal modified MSU-S showed that capacity drastically varied by the order of Al 2 O 3 > CrO 2 > Fe 2 O 3 > Cu 2 O > CeO 2 . Finally, it was found that Temkin and pseudo-second order models are the best models to define DBT adsorption on Al 2 O 3 -MSU-S equilibrium and kinetics phenomena.