Kinetic Modeling of Benzothiophene Catalytic Conversion Over a H

Citing Articles; Related Content. Citation data is made available by participants in Crossref's Cited-by Linking service. For a more comprehensive lis...
0 downloads 0 Views 695KB Size
Article pubs.acs.org/IECR

Kinetic Modeling of Benzothiophene Catalytic Conversion Over a H‑ZSM5 Based Catalyst Saad A. Al-Bogami,†,‡ Jesus Moreira,† and Hugo I. de Lasa*,† †

Chemical Reactor Engineering Centre, Department of Chemical and Biochemical Engineering, Faculty of Engineering, University of Western Ontario, London, Ontario, Canada N6A 5B9 ‡ Research & Development Centre, Saudi Aramco Oil Company, P.O. Box 62, Dhahran 31311, Saudi Arabia ABSTRACT: This study reports a mechanistic based kinetics for the catalytic desulfurization of light diesel hydrocarbons using a zeolite catalyst based on H-ZSM5. An original “parallel” heterogeneous kinetic model assuming a Langmuir−Hinshelwood mechanism is developed. It is based on the experimental data obtained in the CREC Riser Simulator, using benzothiophene/ n-dodecane mixtures at mild temperatures (350−450 °C) and short contact times (3, 5, and 7 s). A lump reaction network is proposed involving benzothiophene, coke-on-catalyst, alkyl-benzothiophene, n-dodecane, and n-dodecane cracking products (paraffins, olefins, and aromatics). The adsorption and intrinsic kinetic model parameters are successfully estimated using regression analysis and a rigorous statistical analysis. The estimated parameters with low spans for the 95% confidence interval and low cross correlation coefficients are able to predict well the experimentally observed products. This developed kinetic model is of great value for the scaling-up of the proposed desulfurization downer/riser technology.

1. INTRODUCTION Heightened environmental legislations mandate a very low or near-zero sulfur content in transportation fuels e.g. gasoline and diesel. On the other hand, nowadays, crude oils which are the sources for these fuels, are getting tougher to process in terms of quality.1,2 In particular, high sulfur levels in these crude oils will require more severe operating conditions using the existing desulfurization technologies in order to comply with environmental regulations. For example, high operating temperatures and large hydrogen consumption will limit the use of the classical hydrodesulfurization (HDS) process.3−5 On this basis, there is a need for an economically viable desulfurization technology by which refiners could meet fuels market specifications. Zeolites emerge as a potential catalyst for a desulfurization process eliminating the use of hydrogen. The acidity of these materials as well as their shape selectivity qualifies them for the purpose of these processes. In this respect, there has been intensive research investigating the conversion of middle distillates (gasoline and diesel) sulfur containing compounds. These research studies considered zeolite based catalysts and mild operating conditions of temperatures (350−450 °C) and close to atmospheric pressure. Namely, H-ZSM5 zeolite was tested for the removal of methyl mercaptan and thiophene sulfur species from hydrocarbon mixtures.3−11 In general, H-ZSM5 zeolite was found to selectively remove sulfur compounds utilizing the hydrogen donor component (e.g., paraffin and olefin) that coexists in the reactants mixture. The Langmuir−Hinshelwood type mechanistic kinetic equations were used widely by many researchers to model the kinetics of sulfur species catalytic conversion. The advantage of using this type of kinetics is that it allows for the accounting of competitive and noncompetitive adsorption.1,8,12−16 The CREC research group has been a leading contributor in the area of sulfur removal during the recent years with studies addressing the removal of mercaptans and thiophene with a © 2013 American Chemical Society

ZSM5 catalyst. While this was successful, it is anticipated that a great potential application for this catalytic process will be in the use of H-ZSM5 zeolites for light diesel desulfurization. A recent study reported by Al-Bogami and de Lasa17 showed the feasibility of this type of desulfurization approach. Thus, given the demonstrated importance of this topic, the aim of the current research is to establish a mechanistically based kinetic model for the removal of sulfur compounds in the light diesel fuel fraction. Benzothiophene is proposed to be used to represent sulfur species to be catalytically converted over a H-ZSM5 catalyst in the CREC Riser Simulator at mild conditions. The proposed heterogeneous kinetic model for benzothiophene conversion is based on the observable reaction products species. Model parameters including kinetic and adsorption constants are estimated using a rigorous statistical analysis. It is envisioned that the proposed original kinetic model will provide a tool of great significance for the scaling up of the proposed fluidized bed technology for light diesel desulfurization.

2. EXPERIMENTAL SECTION 2.1. Catalyst Preparation and Characterization. The fluidizable catalyst particles used in the current research were prepared via sol gel method with 30% H-ZSM5 component and the rest being an inactive silica matrix. Standard catalyst characterization techniques, namely, Particle Size distribution (PSD), Apparent Bulk Density (ABD), X-ray Diffraction (XRD), SEM-EDX, Ammonia Temperature Programmed Desorption (NH3-TPD), Pyridine FTIR, and N2 adsorption were used to investigate different properties of the prepared catalyst. Received: Revised: Accepted: Published: 17760

April 16, 2013 September 29, 2013 October 31, 2013 October 31, 2013 dx.doi.org/10.1021/ie401169k | Ind. Eng. Chem. Res. 2013, 52, 17760−17772

Industrial & Engineering Chemistry Research

Article

(10 mL NORM-JECT, Latex free) through a syringe filter (0.2 μm Supor membrane from Pall Corporation, USA). Finally, the total sulfur was measured by an ICP-OES (Varian Vista Pro; CCD Simultaneous, Australia) auto sampler.

There was no noticeable attrition of the prepared fluidizable particles throughout the several reactions tests performed. Materials and preparation methods as well as theory and experimental procedures of various characterization techniques used in the present study are described by Al-Bogami and de Lasa.17 2.2. Reaction System and Experimental Procedure. Thermal and catalytic runs were established using the CREC Riser Simulator.18 With a volume of about 50 cm3 and 1 g catalyst capacity, the CREC Riser Simulator is a bench scale reactor that mimics the industrial FCC unit. The CREC Riser Simulator has been used extensively to study desulfurization reactions over H-ZSM5 zeolite catalysts such as ethyl mercaptan conversion11 and thiophene conversion.8,9 A detailed diagram of various Riser Simulator components, injection, and sampling procedures can be found elsewhere.17,19 This CREC Riser Simulator unit is equipped with a four port valve (4PV), a six port valve (6PV), and a vacuum box (VB). The 4 PV, 6 PV, and VB have key functions in the operation of the CREC Riser Simulator and as will be discussed later on the development of catalytic conversion of benzothiophene experiments. About 0.8 g of catalyst was loaded into the reactor basket, and the reaction system was sealed and leak tested. Then, the catalyst was further calcined at 550 °C under air flow for 25 min. For the kinetic study, benzothiophene was selected to represent sulfur species while n-dodecane for the diesel fraction itself. Mixtures of benzothiophene and n-dodecane (0.16 g) were reacted at (i) around atmospheric pressure, (ii) mild temperatures (350 °C, 375 °C, 400 °C, 425 °C, and 450 °C), (iii) short contact times (3, 5, and 7 s), and (iv) concentrations of benzothiophene between 0 and 6 wt %. The catalyst to oil ratio (cat/oil) was set to 5 and the impeller velocity to 5700 rpm to get a well fluidized bed. In order to burn any amount of coke formed on the catalyst, a regeneration cycle was required after each catalytic run. The catalyst regeneration process was conducted at a reactor temperature of 550 °C for 25 min under air flow. All thermal and catalytic runs were repeated at least three times to secure the reproducibility of the results. Mass balance closures were in the range of ±6%. 2.3. Product Analysis. The gaseous products from the reaction were analyzed in an Agilent GC/MSD system. An Agilent 5973N with a mass selective detector (MSD) was used to identify the reaction products. An Agilent 6890N gas chromatograph (GC) equipped with both a flame photometric detector (FPD) and a flame ionization detector (FID) allowed product quantifications. On the other hand, the coke deposited on the catalyst surface after experiments was measured in wt % using a total organic carbon analyzer (TOC-V) from Mandel with a solid sample module (SSM-5000). A detailed description of this analytical system along with the TOC can be found elsewhere.17 Moreover, the sulfur amount in the coke was determined using ICP (Inductively Coupled Plasma) and following a method developed by Khan et al.20 About 100 mg of a spent catalyst sample was placed in a digestion tube containing 1 mL of HNO3 and 1 mL of HCL. The tube was then sealed with a screw cap and placed into the ultrasonic bath (VWR Scientific Products, USA, Model 75HT; 117 V, 205 W, with an analogue timer 0 to 35 min, and heater 0 to 85 °C) for sonication. The digestion process was performed at 80 °C for 2 h. Following this, the sample solution was decanted and diluted in a 50 mL volumetric flask using deionized water. The diluted solution was then filtered to an ICP tube using a disposable syringe

3. RESULTS AND DISCUSSION 3.1. Catalyst Characterization. Catalyst properties are reported in Table 1. A full discussion of catalyst characterTable 1. Summary of Characteristics of Catalyst Samples Used in the Current Research17 sample APS (μm) ABD (kg/m3) BET SSA (m2/g) Langmuir SSA (m2/g) DFT SSA (m2/g) micro pore volume (cm3/g) total pore volume (cm3/g) median pore diameter (Å) Si/Al total acidity (mmol NH3/g) Brønsted/Lewis acid sites

H-ZSM5

matrix

pellet

≈1 650 411 509 596 0.116 0.219 6.1 17.65 0.788 1.9

n/a n/a 14 17 8 ≈0 0.026 18 0.69 ≈0 ≈0

50.46 1580 141 175 197 0.036 0.114 6.8 5.05 0.211 1.2

ization techniques used and results are reported by Al-Bogami and de Lasa.17 3.2. Thermal Cracking. Thermal cracking of benzothiophene and n-dodecane was found to be negligible under the most severe conditions investigated in the current research. Namely, at 450 °C and 7 s reaction time, n-dodecane thermal conversion was assessed to be 0.35% and 0.56% in pure n-C12 and mixture samples, respectively. On the other hand, no conversion was observed for benzothiophene due to thermal cracking. As a result, it was concluded that the conversion observed during the catalytic runs truly represents the H-ZSM5 catalytic activity. 3.3. Catalytic Runs. The catalytic runs were conducted using a 6 wt % benzothiophene/n-dodecane mixture as a feed in the CREC Riser Simulator. In addition, catalytic experiments were performed using pure n-dodecane. All catalytic runs were developed in the cyclic mode where a catalytic run is followed by a catalyst regeneration run. 3.3.1. Desulfurization Runs. Positive Identification of Product Species. A typical catalytic desulfurization run in the CREC Riser Simulator consists of a number of procedural steps. These various steps are followed with the change in pressure recorded by the pressure transducer. A typical pressure profile in the Riser Simulator using benzothiophene/ n-dodecane mixture is shown in Figure 1. This figure shows that prior to the injection of the reactant into the reactor, the pressure of the reactor was about 15 psia, whereas the vacuum box is kept at low pressure of 3.7 psia. To maintain this difference in pressure, the reactor and VB are isolated by closing a 4PV. At the time of the reactant injection into the CREC Riser Simulator, the reactant rapidly vaporizes, causing an abrupt increase in reactor pressure (A-B section of the pressure profile in Figure 1). Another stage follows the reactant vaporization, whereby reaction takes place and gaseous products are formed, causing an expansion in the system (B−C section of the pressure profile in Figure 1). 17761

dx.doi.org/10.1021/ie401169k | Ind. Eng. Chem. Res. 2013, 52, 17760−17772

Industrial & Engineering Chemistry Research

Article

Figure 2. Total benzothiophene conversion as a function of reaction time. Reaction conditions: T = 350, 375, 400, 425, and 450 °C, cat/oil = 5. Feed composition is 6 wt % benzothiophene/94 wt % n-dodecane or 60000 ppm of benzothiophene.

Figure 1. Typical pressure profile in the CREC Riser Simulator during a catalytic run with a benzothiophene/n-dodecane mixture. Reaction conditions: T = 350 °C, reaction time = 3 s, and cat/oil = 5. Feed composition is 6 wt % benzothiophene/94 wt % n-dodecane or 60000 ppm of benzothiophene.

Once the preset reaction time is completed, the 4PV is automatically switched to connect the reactor and the vacuum box. The initial large difference in pressure between these two chambers causes the evacuation of all gas phase reaction products and all chemical species adsorbed on the catalyst surface from the reactor into the VB of all gas phase reaction products and all chemical species adsorbed on the catalyst surface. Product evacuation, which occurs almost instantaneously due to the significant differences in pressure and volume of the reactor and VB, leads to a sudden drop in the reactor pressure and consequent rapid pressure stabilization in both chambers pressures (C-D section of the pressure profile in Figure 1). Therefore, chemical species with a molecular diameter smaller than the ZSM5 openings are collected from the gas phase sample using the 6PV and directed later to a GC-MS for analysis. On the other hand, all chemical species with a molecular diameter larger than the ZSM5 opening, such is the case of alkylated benzothiophenes, remain entrapped in ZSM5 network. These entrapped chemical species are measured as coke-on-catalyst once the run completed. Thus and as a result of this quick evacuation gas phase and coke species separation can be achieved. This special CREC Riser Simulator reaction and evacuation function allows establishing a catalytic benzothiophene conversion process, with positive identification of all products including alkylated benzothiophenes species remaining as coke. 3.3.2. Conversion of Benzothiophene. Figure 2 reports the benzothiophene total conversion (aggregated conversion which includes gas phase product species and coke) from the mixture samples containing 6 wt % BZT dissolved in n-dodecane.17 Figures 3 and 4, on the other hand, report the distribution of the conversion in the gas phase products as well as the coke deposited on catalyst, respectively. Conversion calculations can be found elsewhere.17 Different conclusions can be drawn from Figures 2, 3, and 4: (a) Benzothiophene total conversion increases progressively with reaction time and temperature, with a maximum of about 47% at the highest temperature and reaction time (450 °C and 7 s);

Figure 3. Conversion of benzothiophene in gas phase products as a function of reaction time. Reaction conditions: T = 350, 375, 400, 425, and 450 °C, cat/oil = 5. Feed composition is 6 wt % benzothiophene/ 94 wt % n-dodecane or 60000 ppm of benzothiophene.

(b) The H-ZSM5 catalyst is able to selectively remove the sulfur containing compound (BZT) as coke with a very little fraction converted in the gas phase products (Figures 3 and 4); (c) Lower temperatures (350 and 375 °C) favor benzothiophene removal as coke with a negligible fraction converted in the gas phase (less than 1.5%); and (d) At higher temperatures (400 °C−450 °C) the gas phase products from benzothiophene conversion are increased. Benzothiophene has a critical molecular diameter of 6 Å21 while compared to the 5−6 Å size of H-ZSM5 pores. In spite of this, benzothiophene can diffuse toward the H-ZSM5 active sites with some hindrance. The upper critical molecular diameter required for chemical species to evolve into the ZSM-5 zeolite framework is 6.6 Å or smaller. This critical diameter corresponds to durene (tetra-methyl-benzene). Durene is the molecule with the largest critical diameter diffusing out of a ZSM-5 zeolite during methanol conversion.22 Furthermore, at higher temperatures, zeolite pores can expand, facilitating the benzothiophene 17762

dx.doi.org/10.1021/ie401169k | Ind. Eng. Chem. Res. 2013, 52, 17760−17772

Industrial & Engineering Chemistry Research

Article

Figure 4. Conversion of benzothiophene in coke as a function of reaction time. Reaction conditions: T = 350, 375, 400, 425, and 450 °C, cat/oil = 5. Feed composition is 6 wt % benzothiophene/94 wt % n-dodecane or 60000 ppm of benzothiophene .

Figure 5. n-Dodecane conversion as a function of reaction time. Reaction conditions: T = 350, 375, 400, 425, and 450 °C, cat/oil = 5. Feed composition is 100 wt % n-dodecane.

diffusional process. It is reported that the effective ZSM-5 catalytic pore size lies between 6.62 Å and 7.27 Å at 300 °C and can increase to 7.64 Å at 370 °C.23 This fact can explain the higher benzothiophene conversion at higher temperatures as reported in Figure 3, although lower adsorption is expected. On the other hand, alkylated benzothiophenes exhibit larger molecular diameters than benzothiophene and therefore these species can experience increased diffusional constraints into the 5−6 Å H-ZSM5 pores. However, as discussed above, at higher temperatures, zeolite pores may expand allowing alkylated benzothiophene species to leave the zeolite framework. This can also explain the higher benzothiophene conversion toward gas phase products observed at 400 °C−450 °C when compared to the results obtained at 350 °C−375 °C (Figure 3). 3.3.3. Conversion of n-Dodecane. At all conditions investigated in the current research, the pure samples displayed higher conversion of n-dodecane than the mixture samples of 6 wt % benzothiophene. Figures 5 and 6 report the n-dodecane conversion in pure and mixture samples, respectively. It can be noted from Figures 5 and 6 that the n-dodecane conversion increases progressively with reaction time. In addition, at 7 s, the difference in n-dodecane conversion between pure and mixture samples was close to 2% with the mixture samples always given a lower conversion value. However, at a lower reaction time (3 s), this difference is less pronounced than at 5 and 7 s. Thus, it can be argued that the presence of benzothiophene in the reactant feed has a detrimental effect on the conversion of n-dodecane. In other words, when present in the feed mixture, benzothiophene reduces n-dodecane cracking, competing effectively for the same acid sites with this being true, in spite of its larger molecular diameter. Benzothiophene has a 6 Å molecular diameter compared to the 4.9 Å diameter of n-dodecane. However, strong adsorption affinity of benzothiophene will allow its coverage to adsorption sites under the unsteady state operation of the CREC Riser Simulator expected as well as in continuous riser or downer units. Furthermore, one can also notice that at the most severe condition considered (450 °C and 7 s), there is a significant difference between benzothiophene and n-dodecane conversions

Figure 6. n-Dodecane conversion as a function of reaction time. Reaction conditions: T = 350, 375, 400, 425, and 450 °C, cat/oil = 5. Feed composition is 6 wt % benzothiophene/94 wt % n-dodecane or 60000 ppm of benzothiophene.

(47% versus 20%). This can be justified given that at higher thermal levels, benzothiophene experiences less diffusional constraints as a consequence of the expected pore expansion in the H-ZSM5 zeolite framework. 3.3.4. Benzothiophene Selective Conversion. In order to investigate the competitive conversion of benzothiophene and n-dodecane, the definition of a selectivity parameter is valuable8 X SBZT = BTZ XC12 (1) where XBZT and XC12 are the benzothiophene and n-dodecane conversions, with these conversions being based on the WBZT and WC12 injected masses, respectively. Having a selectivity parameter higher than 1 indicates that benzothiophene is preferentially converted with respect to dodecane. One should note that the selectivity parameter lumps the combined effects of adsorption and reaction surpassing the value of 2.1 in all cases studied as in Figure 7. Given the significant 17763

dx.doi.org/10.1021/ie401169k | Ind. Eng. Chem. Res. 2013, 52, 17760−17772

Industrial & Engineering Chemistry Research

Article

fraction of olefins in the products can be attributed to the cracking mechanism over zeolites. It is well-known that n-alkane cracking proceeds via a monomolecular or a bimolecular reaction pathway in which alkenes are formed as intermediates.24−26 One should also notice that there is low yield of aromatic hydrocarbons especially in the products of the n-C12/BZT samples. This is of significant value for process conditions leading to low aromatics in diesel fuel given aromatics decreases the cetane number.27 Concerning the catalytic conversion of benzothiophene, the sulfur products are mainly C1−C3 alkyl-benzothiophenes and sulfur in coke. There was no H2S found in the gas phase products. This result, in turn, points toward the difficulty of benzothiophene cracking under the studied conditions. It is important to point out that the same product distribution patterns were observed under all reaction conditions studied (temperatures of 350 to 450 °C and reaction times 3, 5, and 7s). However, at lower temperatures (350 °C−375 °C) and low reaction time (3 s), the quantification of sulfur products in gas phase alkyl-benzothiophenes was not easy due to their low concentrations. 3.4. Effect of Reactants Initial Concentrations on Catalytic Conversion. In order to explore the reaction order of the benzothiophene reaction, experiments with different initial benzothiophene concentrations (0.5, 2, 4, 6, 8 wt %) in n-dodecane were developed at 450 °C, cat/oil = 5, and 7 s reaction time. The results are shown in Figure 8.

Figure 7. Selectivity of benzothiophene as a function of reaction time. Reaction conditions: T = 350, 375, 400, 425, and 450 °C, cat/oil = 5. Feed composition is 6 wt % benzothiophene/94 wt % n-dodecane or 60000 ppm of benzothiophene.

differences in gas-phase concentrations between n-dodecane and benzothiophene, these results confirm that there is either a higher adsorption affinity of benzothiophene with respect to n-dodecane. Alternatively a much higher intrinsic rate of benzothiophene desulfurization while compared to n-dodecane cracking rate can also be postulated. 3.3.5. Product Distribution. Table 2 reports product distributions, coke on catalyst, and sulfur in coke from a catalytic conversion of pure n-dodecane and a 6 wt % benzothiophene/n-dodecane mixture sample at 450 °C and 7 s reaction time. It can be observed in Table 2 that the converted n-dodecane produced mainly light hydrocarbons (C3 and C4). In addition, there is an appreciable amount of C5−C10 paraffins and cycloparaffins in the cracking products. Furthermore, the small Table 2. Product Distribution from the Catalytic Conversion of Pure n-Dodecane and Benzothiophene/n-Dodecane Mixture at 450 °C, Cat/Oil = 5, and 7 s Reaction Time reactant mixture benzothiophene conversion (%) n-C12 conversion (%) products (wt%) C3−C4 C5−C10a C5−C7= aromaticsb C1-benzothiophene (ppm) C2-benzothiophene (ppm) C3-benzothiophene (ppm) unconverted benzothiophene unconverted n-C12 coke (wt %)c total (wt%) sulfur in coke (wt %)d

100 wt % n-C12

n-C12/6 wt % BZT

20.49

47.48 18.52

14.15 1.98 1.54 2.34

79.58 0.40 100

Figure 8. Benzothiophene total conversion versus benzothiophene initial concentration. Reaction conditions: T = 450 °C, cat/oil = 5, and 7 s reaction time. Total amount of hydrocarbon injected: 0.16 g.

13.21 1.89 1.37 1.84 392.32 50.53 107.99 3.55 77.16 0.93 100 0.24

It can be observed in Figure 8 that benzothiophene conversion decreases when increasing its initial concentration in the reactant mixture. In other words, there is a negative impact on the benzothiophene conversion when its initial concentration in the feed increases. This is a direct indication of a negative reaction order for benzothiophene conversion. This important experimental observation will be discussed in more detail in the upcoming sections of this manuscript when deriving the reaction rate expression for benzothiophene. Figure 9 reports the conversion of n-dodecane versus its weight injected in the Riser Simulator. From the same experiments as reported in Figure 8 with a 0.16 g total mass of hydrocarbons injected, Figure 9 was established. This figure

a Include paraffins and cyclo-paraffins. bInclude benzene, toluene, and alkyl-benzenes. cWeight coke/weight catalyst. dWeight sulfur/weight catalyst.

17764

dx.doi.org/10.1021/ie401169k | Ind. Eng. Chem. Res. 2013, 52, 17760−17772

Industrial & Engineering Chemistry Research

Article

(product shape selectivity). In addition, the benzothiophene reaction over H-ZSM5 zeolite may produce bulkier sulfur molecules species, trapped in zeolite cages and likely end up as coke on catalyst, via transalkylation reactions. FTIR analysis showed that benzothiophene interacts with the hydroxyl group of acid zeolites namely, H-ZSM5 and H-MOR.28 Similar findings are reported by Jaimes and de Lasa8 when converting thiophene on a H-ZSM5 zeolite based catalyst. Hence, it is postulated that benzothiophene alkylation reaction over a H-ZSM5 zeolite proceeds via benzothiophene protonation on a Brønsted acid site. Following this initial step, the resulting carbenium ion reacts with an olefin produced by the cracking of the coreactant n-alkane (n-dodecane in this case). This reaction mechanism is shown in Figure 10. Regarding the produced alkyl-benzothiophene, it is hypothesized that these species isomerize through an alkyl shift mechanism resulting in alkyl-benzothiophenes with alkyl groups placed in different positions in the benzothiophene structure.29,30 For instance, an alkyl shift involving product species as shown in Figure 10 (2-alkylbenzothiophene) leads to 3, 5, and 7alkylbenzothiophene. In line with this, it is important to mention that the kinetic model developed in the present study was established based on an applicable reaction mechanism. This applicable reaction mechanism was employed to support a lumping reaction network as discussed in the upcoming section of the manuscript.

Figure 9. n-Dodecane conversion versus the initial weight fraction injected in the CREC Riser Simulator. Reaction conditions: T = 450 °C, cat/oil = 5, and 7 s reaction time. Total amount of hydrocarbon injected: 0.16 g.

reports n-dodecane conversions for several hydrocarbon mixtures with 99.5 wt %, 98 wt %, 96 wt %, 94 wt %, and 92 wt % n-dodecane. This figure shows that n-dodecane conversion is slightly dependent on its initial concentration. The ndodecane conversion deviates however from the well-known first order behavior of n-alkane cracking where the conversion is independent of initial concentration. This important result suggests that as benzothiophene increases in the feed mixture, the n-dodecane cracking is being reduced. This can be thought of as benzothiophene occupying more active catalytic sites. In turn, this observation indicates the necessity of incorporating the benzothiophene initial concentration in the rate expression of n-dodecane as will be discussed in the next sections.

5. KINETIC MODELING 5.1. Kinetic Model Development. An essential step in benzothiophene kinetic modeling is to identify the main product species. One possible approach, due to the significant number of chemical product species is to lump them into groups according to their molecular properties and boiling point range. It is our view that given the kind of chemical species observed (e.g., sulfur containing species, paraffins, and aromatics) a lumping scheme based on the nature of these chemical species and boiling point range is a reasonable approach leading to 4 lumps. This is an advisable number of lumps to get satisfactory kinetic parameter estimations.31 Hence, for the purpose of the current research, the reaction system is simplified by lumping the product species into groups. To accomplish this, two lumps are defined for benzothiophene conversion: (a) SD (alkyl-benzothiophene in the light diesel boiling point range) and (b) SC (sulfur which remains as coke

4. BENZOTHIOPHENE REACTION MECHANISM Regarding benzothiophene reaction mechanism and based on the product distribution results reported above, it can be stated that benzothiophene mainly undergoes alkylation reactions. In fact, alkyl-benzothiophene species present in the gas-phase products have a molecular diameter such that these species can diffuse out of the zeolite pores. However, alkyl-benzothiophene species with larger alkyl groups are formed and trapped in the zeolite pore network due to their larger kinetic diameter

Figure 10. Benzothiophene alkylation mechanism over H-ZSM5 zeolite catalyst. 17765

dx.doi.org/10.1021/ie401169k | Ind. Eng. Chem. Res. 2013, 52, 17760−17772

Industrial & Engineering Chemistry Research

Article

where rSD is the reaction rate for sulfur species in the diesel range with its rate constants kSD, KBZT is the adsorption constant of benzothiophene, KC12 is the adsorption constant of n-dodecane, and CBZT is the concentration of benzothiophene while CC12 is the concentration of n-dodecane. 3. The formation of alkylated benzothiophene species as coke on catalyst (SC) is a result of adsorbed benzothiophene reacting with a larger (C4+) olefin produced by the cracking of adsorbed n-dodecane. The larger (C4+) olefins can undergo condensation reactions forming coke. Using the steady state approximation one can envision proportionality between C4+ olefin and C12 concentrations. Therefore, the rate equation for the formation of SC can be written as

on catalyst). The proposed network for benzothiophene conversion in the current study is shown in Figure 11. In addition, the proposed kinetic model involves n-dodecane conversion, the model compound used to represent the light diesel fraction. In the same way, the products observed in ndodecane cracking can be grouped as follows: (a) nonaromatic hydrocarbons or HC which include paraffins, olefins, and cycloparaffins and (b) aromatic hydrocarbons or Ar which include benzene, toluene, and alkyl-benzenes. The proposed network for n-dodecane conversion in the current study is shown in Figure 12. It is important to highlight

rSC =

that the coke formation on catalyst contributed by n-dodecane cracking is not considered in the proposed kinetic model. This is based on the experimental observation that n-dodecane conversion remains fairly constant during seven consecutive runs without catalyst regeneration. These seven consecutive runs were performed at the most severe reaction conditions considered in the current research (450 °C reaction temperature and 7 s contact time). The selected reaction networks for benzothiophene and ndodecane, along with the experimental data, allow proposing a “parallel” heterogeneous kinetic model including all observable reactant and product species. Moreover, given the significant importance of selective sulfur species adsorption, a Langmuir− Hinshelwood mechanism is considered for the kinetic model development. It is important to emphasize that this type of mechanism is found best to model the competitive adsorption of reactants and products. The main assumptions for the kinetic model are as follows: 1. The surface chemical reaction is the rate limiting step and is considered not to be affected by the reverse reaction step. 2. The formation of alkylated benzothiophene species in the diesel range (SD) is a result of the reaction of adsorbed benzothiophene with a C2−C3 short olefin. The C2−C3 olefins are produced from adsorbed n-dodecane cracking. The C2−C3 olefins can undergo condensation reactions forming coke. Using the steady state approximation, one can envision proportionality between C2−C3 and C12 concentrations. Thus, the rate equation for the formation of SD can be written as

(3)

rHC =

kHCK C12CC12 (1 + KBZTC BZT + K C12CC12)

rAr =

kArK C12CC12 (1 + KBZTC BZT + K C12CC12)

(4)

(5)

where rHC is the reaction rate for nonaromatics hydrocarbon formation, rAr is the reaction rate for aromatics hydrocarbon formation, kHC and kAr are the corresponding rates constants, KC12 is the adsorption constant of n-dodecane, KBZT is the adsorption constant of benzothiophene, and CC12 is the concentration of n-dodecane while CBZT is the concentration of benzothiophene. It is important to incorporate benzothiophene adsorption in the above two equations. This is in order to account for the experimentally observed effect of benzothiophene initial concentration on n-dodecane conversion (Figure 9). Based on all of the above-mentioned steps and assumptions, the rate of consumption of benzothiophene can be summarized as in eq 6 −rBZT =

(k SD + k SC)KBZTC BZTK C12CC12 (1 + KBZTC BZT + K C12CC12)2

(6)

where rBZT is the reaction rate for benzothiophene reaction with its rate constants kSD and kSC, KBZT is the adsorption constant of benzothiophene, KC12 is the adsorption constant of n-dodecane, and CBZT is the concentration of benzothiophene while CC12 is the concentration of n-dodecane.

k SDKBZTC BZTK C12CC12 (1 + KBZTC BZT + K C12CC12)2

(1 + KBZTC BZT + K C12CC12)2

where rSC is the reaction rate for sulfur species ending as coke on catalyst with its rate constants kSC, KBZT is the adsorption constant of benzothiophene, KC12 is the adsorption constant of n-dodecane, and CBZT is the concentration of benzothiophene while CC12 is the concentration of n-dodecane. 4. The formation of nonaromatic hydrocarbons (HC) and aromatic hydrocarbons (Ar) is the result of the cracking of adsorbed n-dodecane. As a result, the rate equations for HC and Ar can be written as

Figure 11. Proposed reaction network for benzothiophene conversion.

rSD =

k SCKBZTC BZTK C12CC12

(2)

Figure 12. Proposed reaction network for n-dodecane cracking. 17766

dx.doi.org/10.1021/ie401169k | Ind. Eng. Chem. Res. 2013, 52, 17760−17772

Industrial & Engineering Chemistry Research

Article

One can assume that n-dodecane has a low adsorption constant, at high concentrations of benzothiophene KBZTCBZT ≫ 1 + KC12CC12. Therefore, the reaction order becomes negative with respect to benzothiophene. This behavior is shown experimentally in Figure 8 where benzothiophene conversion decreases as its initial concentration increases. In other words, the higher the concentration of benzothiophene the slower the reaction takes place. In this respect, it can be stated that benzothiophene inhibits the reaction. Similarly, the rate equation for the cracking of n-dodecane can be written with a first order kinetics in the numerator as −rC12 =

(kHC + kAr)K C12CC12 (1 + KBZTC BZT + K C12CC12)

and

dyBZT

(7)

dt

=

Wc ⎛ MWiVR ⎞ ⎜ ⎟ri VR ⎝ Whc ⎠

dyC12 dt

dySD dt

For Sulfur in Diesel (SD). where νSD = (MWSD)/(MWBZT). dySC dt

WC νSCβ K BZTK C12 k SCyBZT yC12 VR (1 + αKBZTyBZT + βK C12yC12 )2

=

(17)

For Sulfur in Coke (SC). where νSC = (MWSC)/(MWBZT).

(8)

dyHC dt

=

WC νHCK C12 kHCyC12 VR (1 + αKBZTyBZT + βK C12yC12 )

(18)

For Nonaromatic Hydrocarbons (HC). where νHC = (MWHC)/(MWC12). dyAr

(9)

dt

=

WC νArK C12 kAry VR (1 + αKBZTyBZT + βK C12yC12 ) C12

(19)

For Aromatic Hydrocarbons (Ar). where νAr = (MWAr)/ (MWC12), and MWSD, MWSC, MWHC, and MWAr are the average molecular weights for sulfur in diesel species, sulfur in coke, nonaromatic hydrocarbons, and aromatic hydrocarbons, respectively. These molecular weights were determined based on the yield and molecular weight of individual components formed within each class. Furthermore, each kinetic constant ki, can be postulated to change with reactor temperature T, following the Arrhenius type equation

(10)

⎛ −E ⎡ 1 1 ⎤⎞ ki = kio exp⎜⎜ i ⎢ − ⎥⎟⎟ To ⎦⎠ ⎝ R ⎣T

(20)

where Ei represents the energy of activation, R is the universal gas constant, kio is the pre-exponential factor, and To is the centering temperature In the same way and on the basis of adsorption thermodynamics, one can relate the adsorption constant Ki with the reaction temperature T, as

(11)

yBZT Whc MWBZTVR

WC νSDβKBZTK C12 k SDyBZT yC12 VR (1 + αKBZTyBZT + βK C12yC12 )2

=

(16)

For Benzothiophene. Considering this further, C BZT =

− WC K C12 (kHC + kAr)yC12 VR (1 + αKBZTyBZT + βK C12yC12 )

=

(15)

The next step is the substitution of all rate eqs 2, 3, 4, 5, 6, and 7 into the Riser Simulator design eqs 8 (in terms of concentration) and 10 (in terms of weight fraction). As a result, one can obtain the following: −WC (k SD + k SC)KBZTC BZTK C12CC12 dC BZT = dt VR (1 + KBZTC BZT + K C12CC12)2

(14)

where α = (Whc)/(MWBZTVR) and β = (Whc)/(MWC12VR). For n-Dodecane. Similarly, the consumption of n-dodecane in the CREC Riser Simulator can be evaluated in terms of mass fractions using the following equation:

where Whc is the total mass of hydrocarbons injected in the Riser Simulator (g), MWi is the molecular weight of i (g/mol), and VR is the Riser Simulator volume (cm3). Following substitution of eq 9 into eq 8 and after the required algebraic steps, the CREC Riser Simulator design equation is obtained in terms of species weight fractions as

dyi

(13)

−WC βKBZTK C12 VR (1 + αKBZTyBZT + βK C12yC12 )2 × (k SD + k SC)yBZT yC12

yW i hc MWiVR

=

dt

where ri is the reaction rate of i, VR is the Riser Simulator volume (cm3), Wc is the weight of the catalyst loaded (0.8 gcat), Ci is the concentration of i (mol/cm3), and t is time (s). The concentration of any species (Ci) is related to its weight fraction (yi) as Ci =

MWC12VR

One can establish the benzothiophene consumption rate equation in the CREC Riser Simulator in terms of species mass fractions as

where rC12 is the reaction rate for n-dodecane cracking with its rate constants kHC and kAr, KC12 is the adsorption constant of n-dodecane, KBZT is the adsorption constant of benzothiophene, and CC12 is the concentration of n-dodecane while CBZT is the concentration of benzothiophene. It can be observed that when KBZTCBZT ≫ 1 + KC12CC12 the reaction rate displays a negative order with respect to benzothiophene. This, in turn, can explain the deviation from the first order kinetics observed experimentally as reported in Figure 9, where n-dodecane conversion decreases slightly as the benzothiophene initial concentration increases. This fact can justify the benzothiophene adsorption influence on the cracking reaction rate equation. 5.2. Reaction Modeling in the CREC Riser Simulator. Thus, the CREC Riser Simulator design equation can be written as32

dC VR i = rW i c dt

yC12 Whc

CC12 =

(12) 17767

dx.doi.org/10.1021/ie401169k | Ind. Eng. Chem. Res. 2013, 52, 17760−17772

Industrial & Engineering Chemistry Research ⎛ −ΔH ⎡ 1 1 ⎤⎞ i K i = K io exp⎜⎜ ⎢ − ⎥⎟⎟ To ⎦⎠ ⎝ R ⎣T

Article

Table 3. Intrinsic Kinetics Parameters of the Proposed Kinetic Model along with Their 95% Confidence Intervals in the Case of Pure n-C12d

(21)

where ΔHi is the heat of adsorption, Kio is the pre-exponential factor, and To is the centering temperature. 5.3. Kinetic Parameters Estimation. The derived rate equations, as can be noted from the previous section are highly nonlinear with respect to their parameters. In particular, adsorption constants for benzothiophene and n-dodecane (KBZT and KC12) appear both in the numerator and the denominator of each equation. Therefore, the estimation of kinetic parameters has to be developed applying nonlinear least-squares fitting using MATLAB software. Two built-in available subroutines were used: a) LSQCURVEFIT for the minimization of the objective function and b) ODE113 for the numerical integration of differential equations. The evaluation of model parameters was conducted using a total of 270 experimental data points at different temperatures. The reaction rate constants were expressed using an Arrhenius type of temperature dependence as in eq 20 in order to take into account the temperature dependence of the experimental data. Based on these considerations, each rate constant (ki) yields two parameters (kio and Ei). The same principle is true for adsorption constants of both benzothiophene and n-dodecane using eq 21. Therefore, each adsorption constant (KBZT or KC12) will generate two parameters (Kio and ΔHi). These parameters kio, Ei, Kio, and ΔHi are called intrinsic kinetic parameters. The optimization criterion considers that all rate and adsorption constants must be positive, the activation energy for the reaction must be positive, while the heat of adsorption (ΔH) must be negative. This is to be consistent with physical principles.12,33 The optimization criteria used was based on a minimum sum of squares (SSQ) defined as

95% CI

0.707 140.8 3.591 × 10−4 153.2 2.35 −80. 38 129

0.1663 11.94 1.992 × 10−4 52.54 0.657 12.94

[cm3 gcat−1 s−1]. b[KJ/mol]. c[cm3 mol−1], DOF = m − p, where m is the number of experimental data points, and p is the number of model parameters. dTo = centering temperature = 673 K. a

Table 4. Cross-Correlation Coefficients for the Optimized Kinetic Model Parameters of Table 3 kHCo kAro KC12o EHC EAr ΔHC12

kHCo

kAro

KC12o

EHC

EAr

ΔHC12

1.000 −0.565 −0.314 −0.833 −0.292 0.343

1.000 0.454 0.063 0.758 -0.145

1.000 0.236 0.292 -0.763

1.000 0.233 -0.554

1.000 -0.233

1.000

Table 5. Intrinsic Kinetics Parameters of the Proposed Kinetic Model along with Their 95% Confidence Intervals in the Case of n-C12/BZT Mixturesd parameter kSDoa ESDb kSCoa ESCb kHCoa kAroa KBZToc ΔHBZTb DOF

∑ (xi ,exp − xi ,pred)2 i=1

value

kHCoa EHCb kAroa EArb KC12oc ΔHC12b DOF

N

SSQ =

parameter

(22)

value

95% CI

0.02921 113.75 0.79483 105.19 0.71483 3.833 × 10−4 3.154 × 105 −62.589 262

0.00193 10.961 0.06115 7.538 0.06115 7.513 × 10−4 0.519 × 103 6.023

a [cm3 gcat−1 s−1]. b[KJ/mol]. c[cm3 mol−1], DOF = m − p, where m is the number of experimental data points, and p is the number of model parameters. dTo = centering temperature = 673 K.

where xi,exp and xi, pred are the mass fraction percentages of component i obtained experimentally and predicted by the kinetic model, respectively. The 12 model parameters were evaluated in a two-step process as follows: a) Step 1: Data from pure n-C12 cracking experiments were considered. Calculations involved eqs 15, 18, and 19 with α parameter set at zero. Numerical regression yielded six parameters (kHCo, EHC, kAro, EAr, KC12, ΔHC12). Table 3 reports the six kinetic and adsorption parameters calculated. One should note that as shown in Table 4 all cross-correlation coefficients display values smaller than 0.83 in all cases. The goodness of the fitting of the model as given by eqs 15, 18, and 19 to the various products formed during n-C12 cracking is also reported in the Appendix.b) Step 2: Data from the n-C12/BZT mixture experiments were considered. This second step involved the various model equations (eq 14 to eq 19) and the numerical regression on 6 parameters (kHCo, kAro, KC12, EHC, EAr, ΔHBZT) keeping EHC, EAr, KC12, and ΔHC12 constant. Table 5 reports the six kinetic and adsorption parameters calculated. One should note that all cross-correlation coefficients display values smaller than 0.897 values in all cases as in Table 6.

As shown in Tables 3 and 5, the 12 kinetic parameter calculated show reduced and acceptable spans for the 95% confidence intervals (CI). The estimation of kinetic parameters for the aromatic lumps (kAro) involves however a larger confidence interval, as reported in Table 5. This span is, however, reduced for pure n-C12 cracking as shown in Table 3. In any event the kAro estimated parameter remains in both cases close with the magnitude of the kAr parameter being consistent with the observations of others.33 Moreover, the ability of establishing the 12 model parameters as described above is consistent with the DOF (degree of freedom) in this analysis, with 129 for the first step as shown in Table 3 and 262 for the second step as reported in Table 5. Furthermore, the obtained kinetic parameters are in a good agreement with the experimental observations of product distribution. For example, the energy of activation and the preexponential constant for the conversion of BZT leading to sulfur species in the diesel fraction (SD) are 113.7 KJ/mol and 17768

dx.doi.org/10.1021/ie401169k | Ind. Eng. Chem. Res. 2013, 52, 17760−17772

Industrial & Engineering Chemistry Research

Article

Table 6. Cross-Correlation Coefficients for the Kinetic Model Optimized Parameters KSdo KSco KHco KAro KBzto ESd ESc ΔHBzt

KSdo

KSco

KHco

KAro

KBZTo

ESd

EAr

ΔHBzt

1.000 0.061 −0.219 0.002 −0.220 −0.020 0.191 −0.212

1.000 −0.465 0.009 −0.564 0.153 0.897 −0.733

1.000 −0.068 0.694 −0.163 −0.711 0.765

1.000 −0.010 0.040 0.010 0.004

1.000 −0.163 −0.608 0.854

1.000 −0.096 −0.161

1.000 −0.689

1.000

Figure 13. Comparison between experimental and predicted reaction product yields as a function of reaction time at different reaction temperatures with (a) hydrocarbons (HC), (b) aromatics (Ar), (c) sulfur species in diesel range (SD), and (d) sulfur in coke (SC).

2.92 cm3/gcat s, respectively. The same constant for BZT products remaining as coke-on-catalyst (SC) are 105.1 KJ/mol and 79.4 cm3/gcat s. This suggests that there is indeed a favorable path for benzothiophene being trapped as coke-on-catalyst with the resulting kSC/kSD ratio being 126 while assessed at the centering temperature of 400 °C. This result is a confirmation of the higher BZT conversion leading to coke as reported Figure 4. One can also notice in Table 3 that the activation energy for aromatics formed via n-dodecane cracking (EAr) is much higher than that for nonaromatics hydrocarbons (paraffins, olefins, and cyclo-paraffins): EAr = 153.8 KJ/mol versus EHC = 140 KJ/mol. The determined pre-exponential factors were 0.707 cm3/gcat s

and 0.000359 cm3/gcat s, respectively. This leads to intrinsic kinetic constants where kHC is much higher than that of kAr while assessed at the centering temperature of 400 °C. These results confirm that the high selective conversion of n-dodecane cracking toward paraffins, olefins, and cyclo-parrafins with a very low yield of undesirable aromatics. Moreover, the strong selective adsorption of benzothiophene over n-dodecane is quite evident on the basis of the date from Tables 3 and 5. This is true while considering both the intrinsic adsorption constants and the energies of adsorption: a) n-dodecane yields a KC12o much smaller than KBZTo, and b) the adsorption energy (−ΔH) for benzothiophene is −62.58 KJ/mol while 17769

dx.doi.org/10.1021/ie401169k | Ind. Eng. Chem. Res. 2013, 52, 17760−17772

Industrial & Engineering Chemistry Research

Article

compared to the −80.3 KJ/mol for n-dodecane. These two results lead to a very large KBZT/KC12 ratio of 13420 assessed at the centering temperature of 400 °C. In order to validate the obtained kinetic model, one has to establish a comparison between the experimental data and the results predicted by the model. With this goal in mind, reaction product yields obtained experimentally are compared with the yields predicted by the model in the 400−450 °C range. More specifically, Figure 13 reports a comparison of the mass fraction percentages obtained experimentally with those predicted by the model as a function of reaction time for the reaction products, HC, Ar, SD, and SC, respectively. As shown in this figure, it can be concluded that the proposed kinetic model adequately fits the experimental data within the limits of the experimental errors. Moreover, the parity plot of model predictions as compared to the experimental data is illustrated in Figure 14. It can be

(2) Benzothiophene underwent alkylation reacting with formed olefins resulting from n-dodecane cracking. The benzothiophene initial concentration decreased the benzothiophene conversion rate suggesting a negative reaction order. (3) A “parallel” heterogeneous kinetic model, based on a Langmuir−Hinshelwood mechanism, for benzothiophene conversion over the H-ZSM5 catalyst was successfully implemented. This kinetic model was developed using the observable chemical species including both reactants (benzothiophene and n-dodecane) and products (HC, Ar, SD, and SC). (4) Two separate species lump networks were employed for both benzothiophene and n-dodecane conversion reactions. Sulfur species in the diesel range (SD) and sulfur in coke (SC) were involved in the benzothiophene conversion, while nonaromatic hydrocarbons (HC) and aromatic hydrocarbons (Ar) were produced by n-dodecane cracking. (5) The kinetic model parameters were estimated using nonlinear least-squares fit at different reaction temperatures and reaction times. The estimated parameters with the adequate statistical indicators were able to predict the observed product distribution. The proposed kinetic model which is of great value for the scaling up this process was established with a degree of freedom (DOF) of 258 and was found adequate to describe the experimental data with low spans, low cross correlation coefficients, and a 0.993 R2 regression coefficients.



APPENDIX The following parity plots in Figure 15 show the good fitting of the proposed kinetic model for nC12 cracking using eqs 15, 18,

Figure 14. Parity plot showing the model prediction and experimental data for various lump fractions including unconverted nC12 (■), unconverted benzothiophene (○), paraffinic product lump (x), aromatics product lump (Δ), sulfur species in diesel range lump (☆), and (⧫) sulfur in coke (SC).

inferred that the data is not clustered in horizontal or vertical lines. However, the data points are normally distributed. Horizontal bands may be the result of changes in the observed conversion caused by an independent variable which is not included in the kinetic model. On the other hand, vertical lines are an indication of the kinetic model overparameterization.34,35 Furthermore, the correlation coefficient R2 was found to be 0.993. On this basis, it is concluded that the model predicts the experimental data appropriately.

Figure 15. Parity plot showing the model prediction and experimental data for various fractions including unconverted nC12 (■), hydrocarbons (○), and aromatics product lump (▲).

and 19 with the α parameter set at 0. The parameters used in the simulation are reported in Table 3.

6. CONCLUSIONS The following summarizes the conclusions of the proposed ZSM5 catalytic process for removal of sulfur species in the light diesel range: (1) Benzothiophene was consistently removed in all the runs mainly as coke-on-catalyst with a small fraction being converted into alkyl-benzothiophene species (C1−C3 benzothiophene). This demonstrated the applicability of the proposed downer/ riser technology. There was, in addition, a very modest cracking of n-dodecane yielding mainly lighter paraffins hydrocarbons and a small amount of aromatics.



AUTHOR INFORMATION

Corresponding Author

*Phone: 519-661-2144. E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors would like to acknowledge the financial support of the Natural Science and Engineering Research Council of 17770

dx.doi.org/10.1021/ie401169k | Ind. Eng. Chem. Res. 2013, 52, 17760−17772

Industrial & Engineering Chemistry Research

Article

Symbols

Canada (NSERC). The authors also are very appreciative to Saudi Aramco Oil Company for providing the scholarship to Dr. Saad Al-Bogami to develop this research.



Al = aluminum HCL = hydrochloric acid HNO3 = nitric acid H2S = hydrogen sulfide N2 = nitrogen NH3 = ammonia Si = silicon

NOMENCLATURE

Abbreviations

ABD = average bulk density APS = average particle size Ar = aromatics BET = Brunauer, Emmet, and Teller BZT = benzothiophene cat/oil = catalyst to oil ratio CI = confidence interval CREC = chemical reactor engineering center DFT = density functional theory DOF = degree of freedom EDX = energy dispersive X-ray spectroscopy FCC = fluid catalytic cracking FID = flame ionized detector FPD = flame photometric detector FTIR = Fourier transform infrared spectroscopy GC = gas chromatography HC = hydrocarbon HDS = hydrodesulfurization ICP = inductively coupled plasma MSD = mass selective detector n-C12 or C12 = n-dodecane PSD = particle size distribution rpm = rotation per minute SC = sulfur in coke SD = sulfur in diesel SEM = scanning electron microscopy SSA = specific surface area SSM = solid sample module SSQ = minimum sum of squares T = reactor temperature TOC = total organic carbon TPD = temperature programmed desorption VB = vacuum box XRD = X-ray diffraction 4PV = four port valve 6PV = six port valve

Units



Å = angstrom °C = degrees Celsius cm3 = cubic centimeters g = gram kg = kilogram m3 and m2 = cubic meters and square meters mg = milligram mL = milliliter mmol = millimole ppm = part per million psia = pounds per square inch absolute s = seconds μm = micrometer

REFERENCES

(1) Stanislaus, A.; Marafi, A.; Rana, M. S. Recent advances in the science and technology of ultra low sulfur diesel (ULSD) production. Catal. Today 2010, 153, 1−68. (2) Song, C.; Ma, X. New design approaches to ultra-clean diesel fuels by deep desulfurization and deep dearomatization. Appl. Catal., B 2003, 41, 207−238. (3) Paniv, P. M.; Pyshev, S. V; Gaivanovich, V. I.; Lazorko, O. I. Noncatalytic oxidation desulfurization of the kerosene cut. Chem. Technol. Fuels Oils 2006, 42, 159−166. (4) Ali, M.; Almalki, A.; Elali, B.; Martinie, G.; Siddiqui, M. Deep desulphurization of gasoline and diesel fuels using non-hydrogen consuming techniques. Fuel 2006, 85, 1354−1363. (5) Campos-Martin, J. M.; Capel-Sanchez, M. C.; Fierro, J. L. G. Highly efficient deep desulfurization of fuels by chemical oxidation. Green Chem. 2004, 6, 557−562. (6) Yu, S.; Li, W.; Iglesia, E. Desulfurization of thiophene via hydrogen transfer from alkanes on cation-modified H-ZSM5. J. Catal. 1999, 187, 257−261. (7) Yu, S.; Waku, T.; Iglesia, E. Catalytic desulfurization of thiophene on H-ZSM5 using alkanes as co-reactants. Appl. Catal., A 2003, 242, 111−121. (8) Jaimes, L.; de Lasa, H. I. Catalytic conversion of thiophene under mild conditions over a ZSM-5 catalyst. A kinetic model. Ind. Eng. Chem. Res. 2009, 48, 7505−7516. (9) Jaimes, L.; Ferreira, M. L.; de Lasa, H. I. Thiophene conversion under mild conditions over a ZSM-5 catalyst. Chem. Eng. Sci. 2009, 64, 2539−2561. (10) Jaimes, L.; Ferreira, M. L.; de Lasa, H. I.; Tonetto, G. Desulfurization of FCC gasoline: Novel catalytic processes with zeolites. Int. J. Chem. React. Eng. 2008, 6, 1−66. (11) de Lasa, H. I.; Hernandez Enriquez, R.; Tonetto, G. Catalytic desulfurization of gasoline via dehydrosulfidation. Ind. Eng. Chem. Res. 2006, 45, 1291−1299. (12) Kilanowski, D. R.; Gates, B. C. Kinetics of hydrodesulfurization of benzothiophene catalyzed by sulfided Co-Mo/Al2O3. J. Catal. 1980, 62, 70−78. (13) Vrinat, M. L. The kinetics of the hydrodesulfurization process A review. Appl. Catal. 1983, 6, 137−158. (14) Ho, T. C.; Sobel, J. E. Kinetics of dibenzothiophene hydrodesulfurization. J. Catal. 1991, 128, 581−584.

Parameters

Ci = concentration of i Ki = adsorption constant of i ki = kinetic constant of i MWi = molecular weight of i R = universal gas constant ri = reaction rate of i SBZT = adsorption selectivity if benzothiophene VR = reactor volume wt = weight Wc = weight of catalyst Whc = weight of hydrocarbon injected into the reactor Wsample = weight of catalyst sample used for TOC analysis Xi = conversion of i xi,exp = mass fraction of component i obtained experimentally xi,pred = mass fraction of component i predicted by the kinetic model yi = weight fraction of species i ΔHi = heat of adsorption of i 17771

dx.doi.org/10.1021/ie401169k | Ind. Eng. Chem. Res. 2013, 52, 17760−17772

Industrial & Engineering Chemistry Research

Article

(15) Laredo, G. C.; De los Reyes, J. A.; Luis Cano, J.; Jesús Castillo, J. Inhibition effects of nitrogen compounds on the hydrodesulfurization of dibenzothiophene. Appl. Catal., A 2001, 207, 103−112. (16) Laredo, G. C.; Altamirano, E.; De los Reyes, J. A. Inhibition effects of nitrogen compounds on the hydrodesulfurization of dibenzothiophene: Part 2. Appl. Catal., A 2003, 243, 207−214. (17) Al-Bogami, S. A.; de Lasa, H. I. Catalytic conversion of benzothiophene over a H-ZSM5 based catalyst. Fuel 2013, 108, 490− 501. (18) de Lasa, H. I. Riser simulator U.S. Patent 5 102 628 1992. (19) Kraemer, D. W. Modeling Catalytic Cracking in a Novel Riser Simulator, Ph.D. Thesis, Western University, London, ON, Canada, 1991. (20) Khan, A. H.; Shang, J. Q.; Alam, R. Ultrasound-assisted extraction for total sulphur measurement in mine tailings. J. Hazard. Mater. 2012, 235−236, 376−83. (21) Contreras, R.; Cuevasgarcia, R.; Ramirez, J.; Ruizazuara, L.; Gutierrezalejandre, A; Puentelee, I.; Castillovillalon, P.; Salcedoluna, C. Transformation of thiophene, benzothiophene and dibenzothiophene over Pt/HMFI, Pt/HMOR and Pt/HFAU: Effect of reactant molecular dimensions and zeolite pore diameter over catalyst activity. Catal. Today 2008, 130, 320−326. (22) Ravella, A.; de Lasa, H. I.; Mahay, A. Operation and testing of a novel catalytic reactor configuration for the conversion of methanol to hydrocarbons. Ind. Eng. Chem. Res. 1987, 26, 2546−2552. (23) Van Borm, R.; Aerts, A.; Reyniers, M.-F.; Martens, J. A.; Marin, G. B. Catalytic cracking of 2, 2, 4-trimethylpentane on FAU, MFI, and bimodal porous materials: Influence of Acid properties and pore topology. Ind. Eng. Chem. Res. 2010, 49, 6815−6823. (24) Guisnet, M.; Magnoux, P. Coking and deactivation of zeolites influence of the pore structure. Appl. Catal. 1989, 54, 1−27. (25) Lugstein, A.; Jentys, A.; Vinek, H. n-Heptane cracking on Hand Ni-containing zeolites. J. Chem. Soc., Faraday Trans. 1997, 93, 1837−1842. (26) Ono, Y.; Kanae, K. Transformation of butanes over ZSM-5 zeolites. J. Chem. Soc., Faraday Trans. 1991, 87, 663−667. (27) Cooper, B. H.; Donnis, B. B. L. Aromatic saturation of distillates: An overview. Appl. Catal., A 1996, 137, 203−223. (28) Gutiérrez-Alejandre, A.; Larrubia, M. A.; Ramirez, J.; Busca, G. FT-IR evidence of the interaction of benzothiophene with the hydroxyl groups of H-MFI and H-MOR zeolites. Vib. Spectrosc. 2006, 41, 42−47. (29) Richard, F.; Boita, T.; Moreau, M.; Bachmann, C.; Pérot, G. Transformation of thiophenic compounds over HY zeolite: study of the acid-catalyzed isomerization and disproportionation mechanisms by quantum chemical calculations. J. Mol. Catal. A: Chem. 2007, 273, 48−54. (30) Valla, J. A.; Mouriki, E.; Lappas, A. A.; Vasalos, I. A. The effect of heavy aromatic sulfur compounds on sulfur in cracked naphtha. Catal. Today 2007, 127, 92−98. (31) Valla, J.; Lappas, A.; Vasalos, I. Catalytic cracking of thiophene and benzothiophene: Mechanism and kinetics. Appl. Catal., A 2006, 297, 90−101. (32) Pekediz, A. Oxidative Coupling of Methane, Staged Reactant Feeding in a Riser Simulator, Ph.D. Thesis, University of Western Ontario, London, ON, Canada, 1993. (33) Rahman, F.; Loughlin, K. F.; Al-Saleh, M. A.; Saeed, M. R.; Tukur, N. M.; Hossain, M. M.; Karim, K.; Mamedov, A. Kinetics and mechanism of partial oxidation of ethane to ethylene and acetic acid over MoV type catalysts. Appl. Catal., A 2010, 375, 17−25. (34) El Solh, T.; Jarosch, K.; de Lasa, H. Catalytic dry reforming of methane in a CREC riser simulator kinetic modeling and model discrimination. Ind. Eng. Chem. Res. 2003, 42, 2507−2515. (35) Moreira, J.; Serrano, B.; Ortiz, A.; de Lasa, H. A unified kinetic model for phenol photocatalytic degradation over TiO2 photocatalysts. Chem. Eng. Sci. 2012, 78, 186−203.

17772

dx.doi.org/10.1021/ie401169k | Ind. Eng. Chem. Res. 2013, 52, 17760−17772