A further catalysis mechanism study on Amberlyst 35 resins application in alkylation desulfurization of gasoline

Author's Accepted Manuscript

A further catalysis mechanism study on Amberlyst 35 resins application in alkylation desulfurization of gasoline Rong Wang, Jinbao Wan, Yonghong Li, Hongwei Sun

www.elsevier.com/locate/ces

PII: DOI: Reference:

S0009-2509(15)00392-9 http://dx.doi.org/10.1016/j.ces.2015.05.052 CES12387

To appear in:

Chemical Engineering Science

Received date: 28 November 2014 Revised date: 31 March 2015 Accepted date: 10 May 2015 Cite this article as: Rong Wang, Jinbao Wan, Yonghong Li, Hongwei Sun, A further catalysis mechanism study on Amberlyst 35 resins application in alkylation desulfurization of gasoline, Chemical Engineering Science, http://dx.doi. org/10.1016/j.ces.2015.05.052 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

A further catalysis mechanism study on Amberlyst 35 resins application in alkylation desulfurization of gasoline Rong Wang a,∗, Jinbao Wan a, Yonghong Li b, Hongwei Sun c a

Key Laboratory of Poyang Lake Environment and Resource Utilization, School of Environmental and Chemistry Engineering, Nanchang University, Nanchang 330031, P. R. China;

b

Key Laboratory for Green Chemical Technology, School of Chemical Engineering &Technology, Tianjin University, Tianjin 300072, P. R. China;

c

Department of Chemistry, Nankai University, Tianjin 300071, P. R. China

Abstract Olefinic alkylation of thiophenic sulfurs (OATS) technology can be handled under mild conditions without any hydrogen consumption, which is a promising way to produce clean gasoline. However, the side reaction in this desulfurization process would lead to significant levels of coke. To maintain the high activity for alkylation of sulfurs, the selectivity of OATS catalyst must be improved by reducing side reactions. In this paper, the catalytic mechanism of macroporous sulfonic resins Amberlyst 35 (A35) in the OATS process was further investigated by the calculation of density functional theory (DFT), to understand the reaction path of different reactant at molecular level. The calculated results indicated that the beginning of main and side reactions were both from a stable alkoxide intermediate, which was the protonation

∗

Author for correspondence: School of Environmental and Chemistry Engineering,

Nanchang University, Nanchang 330031, P. R. China. Tel. & Fax: 86-791-83969985 & 83969583, E-mail: [email protected] (R. Wang). 1

product of adsorbed olefin on the catalyst. Compared with alkenes, thiophenic compounds were more inclined to be coadsorbed on the alkoxide intermediate for further reaction, and the alkylation rate of sulfurs with alkenes was faster than the self-dimerization of alkenes. Moreover, the calculated results also indicated that the alkylation of thiophenic sulfurs as main reaction was exothermic while the dimerization of alkenes as side reaction was endothermic over A35. Additionally, the conversion curves of different reactants over A35 and the related kinetics at different temperature were also studied by experimental methods. The obtained experimental results could be used to verify the reliability of relevant theoretical calculations. Based on the differences in the reaction mechanisms obtained by the theoretical and experimental study, two measures were proposed, which would be useful to reduce side reactions to a lower extent. The study would be beneficial to the further industrial application of A35 in the alkylation desulfurization of gasoline. Keywords: Alkylation, Desulfurization, Gasoline, Mechanism, Amberlyst resins 1. Introduction To protect the environment against SOx pollution, the demand for lower sulfur content in transportation fuels has become more and more stringent. It is well known that most of the sulfur present in gasoline (more than 90%) comes from the fluid catalytic cracking (FCC) naphtha cut, which contributes about a third of the gasoline consumed in western Europe and more than 80 vol.% of the total gasoline pool in eastern China. So the aim of gasoline desulfurization is mainly to remove sulfides in the FCC naphtha cut, moreover, thiophene and its derivatives are the major sulfur

2

organic molecules likely to be present in the FCC gasoline (Obame et al., 2013; Babich et al., 2003; Chen et al., 2009). As one of approaches to deep desulfurization for ultra-clean gasoline, the olefinic alkylation of thiophenic sulfur (OATS) process based on increasing the molecular weight of sulfur compounds by alkylation with olefins present in the feed over solid acid catalysts and then followed by distillation, can be handled under relatively mild conditions with a minimal loss of octane number and without any hydrogen consumption. So it can be seen as a good alternative to the conventional catalytic hydro-desulfurization process (Brunet et al., 2005; Dupuy et al., 2012; Jaimes et al., 2009). A series of solid acid catalysts used in the OATS process, such as macroporous sulfonic resins (Amberlyst 35, CT-175 and NKC-9), solid phosphoric acid supported on MCM-41 or Silicalite-1, and modified zeolites (H
, HY, ZSM-5 and HMCM-22) have been examined by several investigations (Dupuy et al., 2012; Jaimes et al., 2009; Wang and Li, 2010; Zhang et al., 2008; Arias et al., 2008; Zheng et al., 2009; Guo et al., 2011; Wang et al., 2012; Bellière et al., 2004; Jaimes et al., 2009). However, the published literature proved that the efficiency of OATS process is still limited by the two side reactions: aromatic alkylation and alkenes oligomerization (Babich et al., 2003). Although the aromatic alkylation could be ignored under the same investigated conditions, the alkenes oligomerization could not be neglected, as there were greatly more of alkenes than thiophenic sulfur compounds in FCC gasoline (Wang and Li, 2010; Zhang et al., 2008; Arias et al., 2008; Zheng et al., 2009). Moreover, the alkenes oligomerization would produce undesired side products with high boiling points,

3

which can cause the deactivation by adsorbing on active sites and plugging catalyst pores. That is harmful to the catalyst stability (Wang and Li, 2010; Zhang et al., 2008; Arias et al., 2008; Zheng et al., 2009; Guo et al., 2011; Wang et al., 2012; Bellière et al., 2004). And too much olefin oligomerization would lead to a decline in the yield of desulfurization

product

in

distillation

column.

Therefore,

to

realize

the

industrialization of OATS technology in FCC gasoline desulfurization, the catalyst selectivity must be improved. How to enhance the product yield and how to prolong the catalyst life by improving the catalyst selectivity, without affecting its highly catalytic activity for sulfur compounds alkylation, is the first and foremost problem. Ambelyst 35 (A35) is a kind of macroporous sulfonic resin with industrialized production and inexpensive price, which is beneficial to the large-scale application. The previous researches proved that A35 had an excellent catalytic activity for thiophenic sulfurs alkylation under a relatively wide and low temperature range (353~383 K) in the OATC process of real FCC gasoline, compared with other OATS catalysts such as solid phosphoric acid, NKC-9 and HY zeolite (Wang and Li, 2010; Guo et al., 2011; Wang et al., 2012). In addition, the resistant temperature of A35 was 413.15 K, and it was the highest among the investigated resins (NKC-9 and D005) (Guo et al., 2010). The A35 resins with the superiority in reaction temperature would be potentially applied in the reactive distillation (RD) of alkylation desulfurization. As the RD process combines the reaction and separation in a single vessel, as well as the gasoline alkylation process consists of catalytic reactions followed by distillation, RD may be a good alternative to reduce the energy consumption for distillation in OATS

4

process. Although A35 has many advantages to be used as OATS catalyst, the side reactions of alkenes oligomerization over A35 are relatively abounding compared to over MCM-41 supported phosphoric acid catalyst, which are adverse to the catalytic stability of A35 (Guo et al., 2010). Given that the mechanisms involving in the reactions on the solid acid catalyst are closely relevant to the activity and selectivity of the catalysts, an improved understanding of how reactant molecules such as thiophenic sulfurs and alkenes interact with acid sites of A35 resins is very beneficial to solve the deactivation problem of A35. As noted by the published study (Namuangruk et al., 2005), theoretical investigations can provide a practical means to elucidate the reaction mechanism at the molecular level, which can be used as the guidance to optimize the reaction conditions for reducing side reactions, to develop novel OATS catalysts with superior performance, as well as to prolong the catalyst lifetime. However, there are little literatures available on understanding the mechanism of major reactions in the OATS process over the A35 resins by means of quantum chemistry calculation (Dupuy et al., 2012; Wang et al., 2011; Liu et al., 2013), except the surface adsorption properties of the catalyst was deeply investigated by the density functional theory (DFT) method of quantum chemistry in our previous study (Wang et al., 2014). Consequently, in the present work, a systematically theoretical calculation was conducted to further gain insights into the reaction mechanism over A35 resins during the alkylation desulfurization process. The geometric optimizations as well as the stationary point characterizations were similarly carried out by DFT calculation, as

5

the DFT method was very useful for elucidating the mechanism of reactions involving solid acid catalysts (Nicholas et al., 1997; Pereira et al., 2006). The theoretical research would reveal the adsorption behavior and reaction pathway of different reactant on the catalyst surface, and the catalytic role of the acidic proton on active sites of A35. Moreover, the effect of temperature on conversions of represented reactants and related kinetics over A35 in the desulfurization process were also investigated by designing experiments in simulated gasoline. The reliability of relevant theoretical calculation could be verified by the experimental data. The combination of theoretical and experimental results was advantageous to the further industrial application of A35 resins in the OATS process. 2. Materials and methods 2.1 Computational Details of DFT On the basis of previous results in simulated as well as real gasoline system (Wang et al., 2014), the aromatics compounds had little effect on the alkylation of thiophenic sulfurs, and the alkenes oligomerization was the major side reaction in the OATS process over A35. Moreover, the most representative sulfide and olefin in FCC gasoline were thiophene (T) and 2-methyl-2-butene (2M2B), respectively. Therefore, T and 2M2B were selected as model reactants for the quantum chemical calculations, and the mono-alkylation of T with 2M2B as well as self-dimerization of 2M2B (Eq. (1) and (2)) were selected as the main and side reaction, respectively, to investigate the catalytic mechanism concisely and effectively. Details of the mechanism were presented herein with all stables, intermediates and transition states computed.

6

S

2

H3 C H3 C

(T)

+

H3C

CH3

H3C

H

(2M2B)

CH3

CH3 H

S H3C

(2M2B)

H3 C

CH3

CH3 (monoalkylated T) CH3 CH3

CH3

C CH3 H2

CH3

+

H2 C

(1)

CH3

CH CH3

C CH3 H2

CH3

(2)

(2,3,4,4-tetramethyl-2-hexene, little) (2,3,4,4-tetramethyl-1-hexene, more)

As one kind of Amberlyst series macroporous sulfonic resins in hydrogen form, A35 resin is prepared by sulfonation of polystyrene cross linked with divinyl benzene. So the catalyst could be modeled by benzene sulfonic acid (Fig. 1a) for simplicity and cost-effectiveness, as it shares the same functional group as A35. The sulfonic acid group in the benzene sulfonic acid is the active group. Although the aromatic ring in the resin has no catalytic activity, the aromatic ring is the electron-donating group and can affect the sulfonic acid group connected on it. The model cluster may not accurately represent the environment and the density of acid sites on the surface of catalyst. However, it is particularly suited to describe local phenomena such as the interaction of organic molecules with active sites and the forming or breaking of bonds by high-qualitative theoretical methods. Calculations with such clusters have proven them to be adequate for qualitative descriptions of chemical rearrangement that occur locally on active sites (Ma et al., 2006; Hansen et al., 2008). Furthermore, our purpose was to study the adsorption property of A35 as well as the reaction pathway of major reactions over the catalyst. All the investigated energies, for example adsorption and activation energies are relative and the cluster model can make the reasonable comparison for them. All the geometric structures investigated in this work was optimized at the 7

B3LYP/6-31+G (d, p) level and the ultrafine integration grid was used to ensure convergence, then the single point energies of the optimized geometries were calculated with the MP2/6-311+G (d, p) level to obtain more realistic values (Svelle et al., 2004; Sousa et al., 2004). Vibrational frequencies were calculated at the same level of optimization to identify the nature of stationary points and obtain zero-point-energy (ZPE) corrections. The reaction trajectory was determined by the intrinsic reaction coordinate (IRC) method to verify the correctness of the obtained theoretical structure of transition state. Gaussian 03 programs were applied to all the calculations and all the calculation results were hold for zero Kelvin (Frisch et al., 2004). 2.2 Catalysts and Materials The compositions of simulated FCC gasoline were listed in Table 1. According to the component analysis, thiophene, C5-olefins with branch chain and toluene were the most representative of sulfides, olefins and aromatics in the FCC gasoline, respectively. Hence thiophene (T), isopentene (ISP) and toluene were selected to simulate gasoline. All the reagents were purchased from Kewei Reagent Company of Tianjin University. Amberlyst 35 (A35) resins used as the alkylation catalyst were purchased from company of Rohm & Hass. The resin had the following characteristics: Exchange capacity (mmol/g [H+])  5.2; Density (g/ml): 0.39-0.42; Specific area (m2/g): 45;

Particle size (mm): 0.70-0.95;

Average porosity size (nm): 24; Max temperature (K): 423.

8

The catalyst needed to be dried at 353 K for 24 h to remove the water from the catalyst structure before the experiments. 2.3 Equipment and Measurements for Catalytic Experiments The tests for the effect of temperature on conversion curves of different reactants catalyzed by A35 during alkylation desulfurization of simulated gasoline were carried out in a 100 mL stirred batch reactor under autogenous pressure. In each run, the autoclave was charged with 4.5 g catalyst and 90 mL gasoline feed with a stirring rate of 600 rpm, which was sufficient to avoid the external mass transfer resistance. When the set reaction temperature was reached, samples were periodically taken out at different reaction time. The volume of each sample was less than 0.05 ml, so the influence on weight ratio of catalyst to gasoline could be ignored. Finally, all the samples were analyzed to obtain conversion curves at different reaction temperatures. Analyses were performed on a FULI 9790 gas chromatograph (Zhejiang, China) that is equipped with two detectors (FID, OV-101 column: 30 m×0.25 mm×0.50 m; FPD, OV-101 column: 60 m×0.25 mm×0.50 m). Meanwhile, different species of sulfides and hydrocarbons in samples were identified by the analysis of a GC-SCD (sulfur compound detector) with a HP-1 column (30 m×0.32 mm×0.25 m) and a GC-MS (mass spectrometer) with a HP5-MS column (30 m×0.25 mm×0.25 m), respectively. 3. Results and Discussion 3.1 DFT Studies on Catalytic Mechanism of A35 for alkylation desulfurization 3.1.1 Theoretical Investigation on Reactants Adsorption on A35 cluster

9

The reactants adsorption on the catalyst surface was the first step in reactions catalyzed by solid catalysts, and the adsorption might be the controlled step of reaction, so the adsorption of representative reactants on A35 cluster was investigated firstly. Considering that the related content had been studied in the previous work about the effect of methanol on catalytic behavior of A35 resins for alkylation desulfurization of FCC gasoline (Wang et al., 2014), the adsorption results were summarized in Fig. 1. Fig. 1 revealed that both 2M2B and T could be physisorbed on the acid site by forming -complexes. Moreover, the weakly adsorbed 2M2B could be further protonated by the acidic proton H1 on the A35 cluster and transformed into a stable alkoxide intermediate (-complex of 2M2B) by forming the covalent bond. The adsorption energy

Eads was calculated by Eq. (3), where Eadsorbate/A35

represents the calculated energy of the adsorbate on the A35 cluster. EA35 and Eadsorbate represent the energies of the separated A35 cluster and the adsorbate, respectively.

!!!!!!!!!!!!!!

Eads = Eadsorbate/A35 – EA35 - Eadsorbate

Based on Equation (3), the lower the

(3)

Eads value is, the stronger the adsorption

strength will be. So 2M2B could be preferentially physically adsorbed on the A35 cluster at the initial stage, as the calculated adsorption energy of -complex of 2M2B was a little lower than that of T 8 kJ/mol (Fig. 1). Moreover, the content of thiophenic compounds was tiny (0.034 wt %) compared with that of alkenes in real FCC gasoline (Wang et al., 2010). It would be considered that almost all the acidic active sites on the catalyst were occupied by 2M2B in the form of -complexes at the initial stage, and then the physisorption of 2M2B transformed into a stable alkoxide intermediate

10

on A35 (-complex of 2M2B) by adsorbing enough energy. As shown in Fig. 1, the calculated activation energy Eact for the transformation was 110 kJ/mol. Therefore, at the end of the adsorption process, most of active sites on the A35 resin were occupied by the stable alkoxide intermediate of 2M2B. 3.1.2 Theoretical Investigation on Reactions over A35 cluster There have been proposed two kinds of reaction mechanisms for alkylation over solid acid catalysts in previous studies, such as the mechanisms of stepwise and concerted (Hansen et al., 2008; Svelle et al., 2004). The stepwise mechanism involves an alkoxide complex formed by the protonation of alkylating reagent on acid sites in the first step, and then the alkoxide complex reacts with another reactant molecule to form an alkylated product. The concerted mechanism involves simultaneous protonation and reaction of two reactant molecules to give the alkylation in one step. The only difference in the two mechanisms is whether or not a stabilized alkoxide intermediate is formed during the reaction course. Our calculation about the catalytic mechanism of A35 for alkylation of T with 2M2B and the dimerization of 2M2B indicated that the alkylated products were formed only after 2M2B interacted with acid sites on A35. As the opposite scenario, the concerted mechanism was also evaluated. While the configurations could not be ruled out on the basis of energies, a physically meaningful transition state for the formation of alkylated product could not be obtained in this case. Moreover, the above calculation for adsorption implied that almost all the acid sites on A35 were occupied by 2M2B in form of -complexes after the adsorption process. Therefore, both the mechanisms of alkylation and

11

dimerization were investigated from a stable alkoxide intermediate of 2M2B on the A35 cluster (-complex of 2M2B, denoted as alkylR1) at the beginning of reaction stage. For comparison, the calculated energy profiles for the alkylation of T with 2M2B as well as the dimerization of 2M2B were both displayed in Fig. 2. The optimized structures of involved stationary points in the alkylation and dimerization reaction were shown in Fig. 3 and 4, respectively. Correspondingly, the major structural parameters of geometric structures in the two reactions were listed in Table 2 and 3. The calculated results about the alkylation in Fig. 2a, Fig. 3 and Table 2 revealed that the further alkylation reaction required the coadsorption of another reactant (T) to the -complex of 2M2B to form a larger complex (coadsT on alkylR1 in Fig. 3a), as the total energy of system was just reduced about -25 kJ/mol after the coadsorption of T. During the coadsorption process, T was diffused into the vicinity of the alkoxide group and formed a weak interaction with it. In spite of the covalent bond O3-C3 on alkylR1 increased by 0.0012 nm, the coadsorption had little effect on the geometric structures. After the reactant T was adsorbed on alkylR1, the alkylation of adsorbed T with alkylR1 started by getting enough energy. The alkylation involved the forming of a covalent bond C6-C3 between the carbon atoms on T and alkylR1, and the proton giving back to the adjacent oxygen atom O1 on the A35 cluster by breaking the proton H3 on the electron-rich site of T ( site). The whole process was clearly displayed by the vibrational motion corresponding to the single imaginary frequency (-221 cm-1) at the transition state for alkylation (TS1) in Fig. 3b. During the transformation, the

12

covalent bond O3-C3 on alkylR1 was broken by stretching to form a carbenium ion transition state, which was unstable and continually attacked the adsorbed T until the bond C3-C6 between the carbenium and T began to form, and the proton H3 of T left toward the active site on A35 cluster The calculated activation energy (Eact) for the alkylation process was 174 kJ/mol. Afterward, the adsorbed alkyl-T on the A35 cluster (Fig. 3c) was desorbed endothermically, which needed the desorption energy of at least 16 kJ/mol (Fig. 2a). The energy profile displayed in Fig. 2b combined with corresponding structures and geometric parameters in Fig. 4 and Table 3 revealed that the reaction path for dimerization of 2M2B was similar to that for alkylation of T. The other 2M2B molecule was also coadsorbed on the stable alkoxide intermediate and formed a complex (coads2M2B on alkylR1 in Fig. 4a). This complex was more stable than the alkylR1 (Fig. 2b). Moreover, to some degree, the coadsorption of another 2M2B weakened the covalent bond O3-C3 on the alkylR1 as well as the double bond C=C of adsorbed 2M2B, as the bond length was increased of 0.0007 and 0.0002 nm, respectively (Table 3). Then, the reaction proceeded by breaking the covalent bond between the alkylR1 and the catalyst sites, and forming a new C-C bond to the other 2M2B molecule. An unstable isopentyl carbenium ion was also found in the transition state (TS2) for self-dimerization of 2M2B on A35 (Fig. 4b). The calculation proved that the vibrational motion in TS2 was corresponding to the sole imaginary frequency (-287 cm-1), which described the reaction involved the breaking of bond O3-C3 and formation of bond C3-C8. The structures of TS2 displayed that the covalent bond on

13

the alkylR1 was broken with a large increase from 0.1510 to 0.3495 nm, and the atom of C3 was located about midway between the resin oxygen atom O3 and C8 on the double bond of 2M2B, where a new bond C3-C8 was forming. The similarity of reaction paths in the above calculation indicated that dimerization of alkenes as side reaction during the OATS process catalyzed by A35 could be not completely eliminated. However, there were still two obvious differences between the main and side reactions by analyzing the obtained catalytic mechanism. Firstly, the system energy changes in the reaction process (Fig. 2) demonstrated that the alkylation of T was an exothermic process whereas the dimerization of 2M2B was an endothermic, which was consistent with the thermodynamic calculation results obtained by Bellière et al. (2004). It seemed that if the reaction temperature was too high, the side reactions of alkenes oligomerization would be promoted, which could reduce the catalyst activity for alkylation of sulfurs. Secondly, as shown in Fig. 2, the energy change in coadsorption of 2M2B on alkylR1 was less of 13 kJ/mol than that of T. It revealed that T was more inclined to be coadsorbed on the alkoxide complex than 2M2B. Moreover, the activation energy for alkylation of T was obviously lower of 34 kJ/mol than that for self-dimerization of 2M2B, indicating that the alkylation of thiophenic compounds with olefins was faster than the self-dimerization of alkenes over the A35 cluster. So the alkylation of thiophenic sulfurs happened more easily compared with the self-dimerization of alkenes over the A35 resins. 3.2 Experimental studies during the Alkylation desulfurization process over A35 In order to verify the reliability of the above theoretical calculation, the

14

relationships between conversions of different reactants and reaction time over A35 were investigated at different reaction temperatures by the experimental method. It should be note that the experiments were carried out in simulated gasoline system to avoid the complexity of components in real gasoline. According to the experimental procedures in Section 2.3, the relevant experimental results were shown in Fig. 5. Given that the thiophene concentration in the feed (Table 1) was tiny, the alkenes consumed in the alkylation of thiophene could be neglected and its conversion was considered as the consumption in side reactions, such as alkenes oligomerization and aromatic alkylation. However, no alkyl-product of toluene was found in the products over the investigated temperature range 353-393 K. The effect of alkylation products of aromatics on catalytic stability of A35 could be ignored and the main side reactions were alkenes oligomerization during the OATS process catalyzed by A35. Fig. 5a showed the conversion curves of thiophene under different temperatures, which indicated that increasing temperature was beneficial to improving the catalyst activity for alkylation of T in the suitable temperature range 353-383 K. However, the higher temperature was not always conducive to the alkylation, as a decreasing tendency in the conversion was observed when the temperature was 393 K. It implied that the alkylation of T was favored at a relatively low temperature range (353-383 K). In this range, rising temperature could increase the conversion of T for alkylation, since high temperature could enhance the movement of molecules and promote the adsorption of reactants on the acid sites of A35. However, the catalytic activity of A35 was suppressed at higher temperature 393 K, which meant that the alkylation of

15

sulfurs over A35 should be an exothermic reaction and the catalytic activity was not always increased with increasing temperature. Conversely, the conversion curves of 2M2B in Fig. 5b revealed that a higher temperature was favorable for the reaction of alkenes. The conversion was raised with increasing temperature, and the highest conversion of 2M2B was at 393 K, which revealed that the dimerization of 2M2B over A35 should be an endothermic reaction and a high temperature was advantageous to the occurrence of dimerization of alkenes. These experimental results were conformed to the energy changes in theoretical outcome in Section 3.1. Additionally, the apparent kinetics parameters for the alkylation of T and dimerization of 2M2B were also determined. Considering that the power series model was the simplest form of kinetic models and widely used in many reaction systems, the reaction rate equations of T and 2M2B were expressed by Eq. (4) and (5), respectively, where r is the reaction rate, k is the reaction constant, C (mmol/L) is the concentration at the reaction time of t (h), Į, ȕ and  are power series.

rT = -dCT/dt = kTCT C2M2B


r2M2B= -dC2M2B/dt = k2M2BC2M2B

(4)

(5)

As listed in Table 1, the weight percent of olefins was enormously higher than the content of T. Even if the concentration of olefin was to be reduced to 4.0 wt% in the process of catalyst deactivation, it would still be 100 times higher than the concentration of T. Therefore, the concentration of 2M2B in Eq. (4) could be seen as a constant compared with the tiny consumption for alkylation of sulfides. Moreover, the power series Į and  in Eq. (4) and (5) were firstly hypothesized to be equal to 1. According to the above hypothesis, Eq. (6) and (7) were obtained by simplifying, in

16

which r, k and C are the same significations as in the Eq. (4) and (5).

rT = -dCT/dt = kTCT

r2M2B= -dC2M2B/dt = k2M2BC2M2B

(6)

(7)

Then, Eq. (8) could be obtained by the integration of Eq. (6) and (7). Based on the experiment datum xi-t at different temperatures in Fig. 5, the relations of –ln(1- xi) - t were displayed in Fig. 6 and the plots were nearly linear in all cases, indicating both the apparent kinetic models for the alkylation of T and dimerization of 2M2B might be first-order reactions. The reaction rate constants Ki could be obtained from the slopes of these plots. The Ki values of different reactants under various reaction temperatures were presented in Table 4. The relational graphs of lnKi vs 1/T(K)×103 were shown in Fig. 7, demonstrating that the obtained results conformed to the Arrhenius Law (Eq. (9), in which A is pre-exponential factor and Eact is activation energy). ln(Ci0/ Ci)= -ln(1- xi) = Kit (i=T or 2M2B) Ki = A i exp (-Eact, i/RT)

(8)

or lnKi = lnA i - Eact, i/RT (i=T or 2M2B)

(9)

In conclusion, the apparent kinetic equations for the alkylation of T with 2M2B and the dimerization of 2M2B catalyzed by A35 could be expressed by the following Eq. (10) and (11), respectively.

rT(mmol/Lh) = KTCT , r2M2B(mmol/Lh) = K2M2BC2M2B ,

KT = 1.67×106exp(-4.94×103/T )

(10)

K2M2B = 4.86×107exp(-6.29×103/T ) (11)

On the premise of existing hypothesis, the dynamics appearances of main and side reactions were similar. However, as shown in Fig. 7, the order of apparent activity

17

energies (Eact) of the two reactions was: dimerization of 2M2B >alkylation of T, which meant that the side reaction could be increased rapidly by increasing the temperature. So it was necessary to carry out the OATS process under the appropriate reaction temperature, which was beneficial to the alkylation conversion of thiophenic sulfides. As shown in Table 4, all the reaction rate constants of T were higher than that of 2M2B in the studied temperatures range, revealing that the alkylation of T was faster than the dimerization of 2M2B. The initial conversion of thiophene in Fig. 5a (before 0.5 h) was very fast under the investigated temperatures, and there was a tiny increase in the conversion rate when the reaction time was longer than 1 h. However, the conversion rate of 2M2B in Fig. 5b was relatively low before 0.5 h, which might be caused by the competition of thiophene alkylation at the initial stage. This experimental conclusion could be also used to prove the relevant theoretical calculation. According to the calculated results, the alkylation of thiophenic sulfurs with alkenes happened more easily compared with the self-dimerization of alkenes over the cluster of A35, since thiophene was more inclined to be coadsorbed on the activated olefin than 2M2B for the further reaction, and the activation energy for alkylation of thiophene was lower of 34 kJ/mol than that for dimerization of 2M2B. In the end, both the experimental kinetics results as well as the quantum calculation outcomes were concluded and compared in Table 5. Although there was a big difference between in the theoretical and experimental activation energies, which might be caused by the limitation of cluster model used in the theoretical calculation as well as the application of empirical Arrhenius Formula, it has little impact on the

18

analysis of catalytic mechanism as well as the validation of theoretical results by the experimental methods, since the orders of these activation energies were consistent. As shown in Table 5, the consistency of results obtained by theoretical calculation and experimentation could be used to prove the correctness of the theoretical calculation results. It should be noted that the similarity in reaction paths decided that the dimerization of alkenes as side reaction was impossible to be completely eliminated in the OATS process catalyzed by A35. However, the differences in the main and side reaction mechanisms (Table 5) could be utilized to reduce the occurrences of side reaction as low as possible, while the high activity for alkylation desulfurization was kept. Therefore, same measures seemed to be useful to improve the catalytic behavior of A35 in the process were proposed. Firstly, the theoretical results showed that the beginning of main and side reactions were both from a stable alkoxide intermediate (-complex of alkenes), which was the protonation product of adsorbed olefin on the acid sites of A35. Considering that the thiophenic sulfurs content in the feed was so tiny that the demand for activated olefins (-complex of alkenes) to react was small (Guo et al., 2011; Wang et al., 2014), if there were excess active sites on A35 for the formation of activated olefins, the redundant activated olefins would participate in side reactions, which was adverse to increase the catalytic selectivity. Our previous work had proved that methanol and its ether product with olefin could compete with thiophenic sulfurs and alkenes for active sites on the A35 (Wang et al., 2010). The competition of suitable amount of methanol could reduce the amount of adsorbed olefins on acid

19

sites to form the activated alkenes. Moreover, the calculated results demonstrated that sulfurs were more inclined to be coadsorbed on the activated alkenes for the further reaction, and the lower activation energy made the main reaction happen easily. Therefore, adding suitable amount of methanol in the OATS process catalyzed by A35, had little impact on the alkylation of sulfurs, but could decrease the conversion of alkenes for side reaction, by reducing the amount of redundant activated olefins on the surface of A35. So it would be favorable for improving the catalytic selectivity and prolonging the catalyst life. Secondly, the application of reactive distillation in the alkylation desulfurization process over A35 might be the other way to prolong the catalyst service life (Guo et al., 2012). Since the results proved that the alkylation of thiophenic sulfurs was exothermic while the oligomerization of olefins was endothermic, strictly controlling the reaction zone at the desired temperature range not only improved the desulfurization rate, but also minimized the occurrence of alkenes oligomerization. Additionally, the catalytic mechanism revealed that the alkylation of sulfurs over A35 was more easily and faster than that the oligomerization of alkenes. So in the reactive distillation column, the redundant alkenes after the alkylation of sulfurs could be continuously separated from the reaction zone to reduce their residence time on the catalyst and minimize their side effects to a lower extent, while the high desulfurization rate was not affected.

20

4. Conclusions The typically main and side reactions in alkylation desulfurization process of FCC gasoline catalyzed by Amberlyst 35 resins were investigated by the DFT method of quantum chemistry. The theoretical calculation for the reaction paths showed that the beginning of the main and side reaction were both from a stable alkoxide intermediate of olefin (activated olefin), that was formed by protonation of olefin adsorbed on the catalyst. Thiophenic sulfurs were more inclined to be coadsorbed on the alkoxide complex than alkenes for further reaction, and the activation energy for alkylation of sulfurs was obviously lower of 34 kJ/mol than that for dimerization of alkenes. The alkylation rate of thiophenic sulfurs over A35 was faster than the dimerzation rate of alkenes. Moreover, the theoretical results also indicated that alkylation of sulfurs was an exothermic reaction while dimerization of alkenes was an endothermic reaction catalyzed by A35. It could be concluded that the alkylation of thiophenic compounds as the main reaction would occur more easily compared with the side reactions of dimerization of alkenes under suitable reaction conditions. Additionally, the conversion curves of different reactants as well as the related kinetics over A35 under different reaction temperature were also studied by the experimental method. The obtained experimental results could be used to verify the reliability of relevant theoretical calculations. The combination of the theoretical and experimental study could provide reliably theoretical guidances to optimize the reaction conditions for a more efficient desulfurization by reducing side reactions to a lower extent, which would be beneficial to the further industrial application of A35

21

resins in the alkylation desulfurization process of FCC gasoline.

Acknowledgements The authors gratefully acknowledge the financial support provided by National Natural Science Foundation (No. 21206067), China Postdoctoral Science Foundation (No. 2012M511455) and Postdoctoral Foundation for Excellent Scientific Research Project in Jiangxi Province. References Obame, H., Toussaint, G., Laurenti, D., Tayakout, M., Geantet, C., 2013. Comprehensive GC×GC Characterization of the Catalytic Alkylation of Thiophenic Compounds in a FCC Gasoline. Top. Catal. 56, 1731-1739. Babich, I.V., Moulijn, J.A., 2003. Science and technology of novel processes for deep desulfurization of oil refinery streams: a review. Fuel 82, 607-631. Chen, H., Wang, Y.H., Yang F.H., Yang, R.T., 2009. Desulfurization of high-sulfur jet fuel by mesoporous -complexation adsorbents. Chem. Eng. Sci. 64, 5240-5246. Brunet, S., Mey, D., Pérot, G.., Bouchy, C., Diehl, F., 2005. On the hydrodesulfurization of FCC gasoline: a review. Appl. Catal. A 278, 143-172. Dupuy, B., Laforge, S., Bachmann, C., Magnoux, P., Richard, F., 2012. Desulfurization of model FCC feedstocks by alkylation: Transformation of thiophenic compounds in presence of 2-methyl-1-pentene over acidic zeolites. J. Mol. Catal. A 363, 273-282. Jaimes, L., De Lasa, H., 2009. Catalytic Conversion of Thiophene under Mild 22

Conditions over a ZSM-5 Catalyst: A Kinetic Model. Ind. Eng. Chem. Res. 48, 7505-7516. Wang, R., Li, Y.H., 2010. Preparation of MCM-41 supported phosphoric acid catalyst for thiophenic compounds alkylation in FCC gasoline. Catal. Commun. 11, 705-709. Zhang, Z.K., Guo, X.Y., Liu, S.L., Zhu, X.X., Xu L.Y., 2008. Olefins alkylation thiophenic sulfur of real gasoline over the fluorinated H
zeolite catalyst. Fuel Process Technol. 89, 1135-1141. Arias, M., Laurenti, D., Geanete, C., Vrinat, M., Hideyuki, I., Yoshimura, Y., 2008. Gasoline

desulfurization

by

catalytic

alkylation

over

silica-supported

heteropolyacids: From model reaction to real feed conversion. Catal. Today 130, 190-194. Zheng, X.D., Dong, H.J., Wang, X., Shi, L., 2009. Study on olefin alkylation of thiophenic sulfur in FCC gasoline using La2O3-modified HY zeolite. Catal. Lett. 127, 70-74. Guo, B.S., Wang, R., Li, Y.H., 2011. Gasoline alkylation desulfurization over Amberlyst 35 resin: Influence of methanol and apparent reaction kinetics. Fuel 90, 713-718. Wang, R., Li, Y.H., Guo, B.S., Sun, H.W., 2012. Effect of Methanol on Catalytic Performance of HY Zeolite for Desulfurization of FCC Gasoline by Alkylation. Ind. Eng. Chem. Res. 51, 6320-6326. Bellière, V., Geantet, C., Vrinat, M., Ben-Taarit, Y., Yoshimura, Y., 2004. Alkylation

23

of 3-Methylthiophene with 2-Methyl-2-butene over a Zeolitic Catalyst. Energy Fuels 18, 1806-1813. Jaimes, L., Ferreira, M.L., de Lasa, H., 2009. Thiophene conversion under mild conditions over a ZSM-5 catalyst. Chem. Eng. Sci. 64, 2539-2561. Guo, B., Wang, R., Li, Y., 2010. The Performance of Solid Phosphoric Acid Catalysts and Macroporous Sulfonic Resins on GasolineAlkylation Desulfurization. Fuel Process Technol. 91, 1731-1735. Namuangruk, S., Pantu, P., Limtrakul, J., 2005. Investigation of Ethylene Dimerization over Faujasite Zeolite by the ONIOM Method. Chem. Phys. Chem. 6, 1333-1339. Dupuy, B., Laforge, S., Morais, C., Bachmann, C., Magnoux, P., Richard, F., 2012. Alkylation of 3-Methylthiophene by 2-Methyl-1-Pentene over HY, H
and HMCM-22 Acidic Zeolites. Appl. Catal. A 413, 192-204. Wang, R., Li, Y.H., Guo, B.S., Sun, H.W., 2011. Catalytic mechanism of MCM-41 supported Phosphoric acid catalyst for FCC gasoline desulfurizatin by alkylation: experimental and theoretical investigation. Energy Fuels 25, 3940–3949. Liu, Y., Yang, B.L., Yi, C.H., 2013. Density Functional Theory Investigation for Catalytic Mechanism of Gasoline Alkylation Desulfurization over NKC-9 Ion-Exchange Resin. Ind. Eng. Chem. Res. 52, 6933-6940. Wang, R., Wan, J.B., Li, Y.H., Sun, H.W., 2014. An insight into effect of methanol on catalytic behavior of Amberlyst 35 resins for alkylation desulfurization of fluid catalytic cracking gasoline. Fuel 115, 609-617.

24

Nicholas, J.B., 1997. Density Functional Theory Studies of Zeolite Structure, Acidity, and Reactivity. Top. Catal. 4, 157-171. Pereira, M.S., Chaer Nascimento, M.A., 2006. Theoretical Study on the Dehydrogenation Reaction of Alkanes Catalyzed by Zeolites Containing Nonframework Gallium Species. J. Phys. Chem. B 110, 3231-3238. Ma, Q.S., Chakraborty, D., Faglioni, F., Muller, R.P., Goddard ; , W.A., 2006. Alkylation of Phenol: A Mechanistic View. J. Phys. Chem. A 110, 2246-2252. Hansen, N., Brüggemann, T., Bell, A.T., Keil, F.J., 2008. Theoretical Investigation of Benzene Alkylation with Ethene over H-ZSM-5. J. Phys. Chem. C 112, 15402-15411. Svelle, S., Kolboe, S., Swang, O., 2004. Theoretical Investigation of the Dimerization of Linear Alkenes Catalyzed by Acidic Zeolites. J. Phys. Chem. B 108, 2953-2962. Sousa, S.F., Fernandes, P.A., Ramos, M.J., 2007. General Performance of Density Functionals. J. Phys. Chem. A 111, 10439-10452. Frisch, M.J., Trucks, G.W., Schlegel, H.B., Scuseria, G.E., Robb, M.A., Cheeseman, J.R., Montgomery, Jr., J.A., Vreven, T., Kudin, K.N., Burant, J.C., Millam, J.M., Iyengar, S.S., Tomasi, J., Barone, V., Mennucci, B., Cossi, M., Scalmani, G., Rega, N., Petersson, G.A., Nakatsuji, H., Hada, M., Ehara, M., Toyota, K., Fukuda, R., Hasegawa, J., Ishida, M., Nakajima, T., Honda, Y., Kitao, O., Nakai, H., Klene, M., Li, X., Knox, J.E., Hratchian, H.P., Cross, J.B., Adamo, C., Jaramillo, J., Gomperts, R., Stratmann, R.E., Yazyev, O., Austin, A.J., Cammi, R., Pomelli, C.,

25

Ochterski, J.W., Ayala, P.Y., Morokuma, K., Voth, G.A., Salvador, P., Dannenberg, J.J., Zakrzewski, V.G., Dapprich, S., Daniels, A.D., Strain, M.C., Farkas, O., Malick, D.K., Rabuck, A.D., Raghavachari, K., Foresman, J.B., Ortiz, J.V., Cui, Q., Baboul, A.G., Clifford, S., Cioslowski, J., Stefanov, B.B., Liu, G., Liashenko, A., Piskorz, P., Komaromi, I., Martin, R.L., Fox, D.J., Keith, T., Al-Laham, M.A., Peng, C.Y., Nanayakkara, A., Challacombe, M., Gill, P.M.W., Johnson, B., Chen, W., Wong, M.W., Gonzalez, C., Pople, J.A., 2004. Gaussian 03, Revision C.01, Gaussian Inc., Wallingford CT. Guo, B.S., Li, Y.H., 2012. Analysis and simulation of reactive distillation for gasoline alkylation desulfurization. Chem. Eng. Sci. 72, 115-125.

26

Figure captions: Fig. 1. Optimized geometries: (a) the cluster model of A35 resin, (b) the single thiophene adsorbed on the cluster of A35 (-complex of T on A35 cluster), (c) -complex of 2M2B on A35 cluster, (d) transition state (TS) of 2M2B from - to -complex on A35 cluster and its vibrational motion, (e) -complex of 2M2B on A35 cluster. Fig. 2. Calculated energy profiles for the alkylation and dimerization over the A35 cluster: (a) alkylation of T with -complex of 2M2B, (b) dimerization of 2M2B on the A35 cluster. Fig. 3, Optimized geometric structures of different stationary states during the alkylation of T with -complex of 2M2B: (a) coadsorption of T on -complex of 2M2B on A35 cluster (coadsT on alkylR1), (b) transition state for alkylation of T and 2M2B on A35 cluster (TS1) and its vibrational motion, (c) alkyl-T on A35 cluster. Fig. 4. Optimized geometric structures of different stationary states during the self-dimerization of 2M2B: (a) coadsorption of 2M2B on -complex of 2M2B on A35 cluster (coads2M2B on alkylR1), (b) transition state for dimerization of 2M2B on A35 cluster (TS2) and its vibrational motion, (c) dimer-2M2B on A35 cluster. Fig. 5. Conversion curves of the reactants in simulated FCC gasoline at different reaction temperatures: (a) Thiophene; (b) 2-methyl-2-butene (2M2B). Fig. 6. The deduction for kinetics equations of different reactions in the OATS

27

process over A35: (a) alkylation of T with 2M2B; (b) dimerization of 2M2B. Fig. 7. The validation of obtained kinetics equations by relational graph of lnKi vs 1/T(K)×103.

28

Highlights

! The catalysis mechanism of A35 in OATS process is studied by DFT method in depth.

! The

path of main and side reactions over A35 is similar but there is two

differences.

! Based on the differences, two methods are proposed to improve the selectivity of A35.

! The correctness of relevant theoretical calculations can be verified by experiments. Table 1. Compositions of simulated FCC gasoline in experiments Compositions

Contents (wt.%)

Toluene (>99 wt.%) Isopentene (>99 wt.%, containing 7.2 wt.% 2-methyl-1-butene and 92.8 wt.% 2-methyl-2-butene ) n- octane (>99 wt.%) Thiophene (>99 wt.%)

15.3

29

37.4 47.3 0.034

Table 2. Part geometric parameters of adsorption complexes of reactant, transition state (TS) and product during the alkylation of T on A35 (Distances in nm, Angles in Degrees) Alkyl-T on T + alkylR1 coadsT on alkylR1 TS1 A35 cluster O3C3 SO3 SO2 SO1 SC1 O1H3 C3O3S C2C3 C2H1 C6C7 C6H3 C7H3 C3C7 C3C6 S1C6C7

0.1503 0.1630 0.1467 0.1465 0.1792

0.1515 0.1632 0.1469 0.1465 0.1792

126.2 0.1545 0.1098 0.1370 0.1081

126.5 0.1545 0.1098 0.1372 0.1082

0.1501 0.1108 0.1401 0.1105

111.5

0.3438 0.3188 111.7

0.3030 0.2367 109.6

30

0.4035 0.1504 0.1489 0.1502 0.1806 0.1962

0.1462 0.1470 0.1638 0.1788 0.0983 0.1562 0.1098 0.1374 0.3344 0.4171 0.1522 109.1

Table 3. Part geometric parameters of adsorption complexes of reactant, transition state (TS) and product during the dimerization of 2M2B on A35 (Distances in nm, Angles in Degrees) 2M2B + coads 2M2B on Dimer-2M2B on TS2 alkylR1 alkylR1 A35 cluster O3C3 SO3 SO2 SO1 SC1 C3O3S C2C3 C3C8 C3C9 C8C9 C9C10 C10H5 C8H4 O1H5 H4C8C9C10

0.1503 0.1630 0.1467 0.1465 0.1792 126.2 0.1545

0.1344 0.1511 0.1094 0.1091 0

0.1510 0.1629 0.1467 0.1466 0.1793 126.1 0.1545 0.3294 0.3718 0.1346 0.1512 0.1094 0.1091 0.3840 0.2

0.3495 0.1481 0.1469 0.1587 0.1796

0.1465 0.1467 0.1637 0.1789

0.1504 0.2275 0.3065 0.1415 0.1430 0.1318 0.1086 0.1188 6.5

0.1560 0.1588 0.2655 0.1530 0.1348 0.2146 0.1097 0.0990 10.7

Table 4. Reaction rate constant (Ki) of different reactions under vary temperatures over A35 T/K Kalkyl-T/h

-1

Kdimer-2M2B/h-1

353

363

373

383

1.37

2.14

3.03

4.11

0.98

1.58

2.47

3.98

Table 5. Results comparison of quantum calculations and experimental kinetics Quantum calculations

Experimental kinetics

Activation energy (Eact) Eact of alkyl-T = 174 kJ/mol Eact of dimer-2M2B = 208 kJ/mol Eact of alkyl-T < Eact of dimer-2M2B Reaction rate (r)

Eact of alkyl-T = 41 kJ/mol Eact of dimer-2M2B = 52 kJ/mol Eact of alkyl-T < Eact of dimer-2M2B Reaction rate constant of T higher than that of 2M2B at investigated temperatures r of alkyl-T > r of dimer-2M2B

coadsT on alkylR1 faster than that of 2M2B r of alkyl-T > r of dimer-2M2B Thermodynamic property alkyl-T (exothermic) Dimeri-2M2B (endothermic)

alkyl-T (exothermic) Dimeri-2M2B (endothermic)

31

Graphical Abstract (for review)

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7