A Study on the Coking Sites of γ-Alumina Surface Using 1-Methylnaphthalene as the Model Reactant

Hu Anpeng; Han Wei; Zhang Le; Long Xiangyun; Li Mingfeng; Nie Hong

(SINOPEC Research Institute of Petroleum Processing, Beijing 100083)

Abstract: The support γ-Al2O3 was treated with 1-methylnaphthalene as the model reactant by respectively using the chemical static adsorption method and the accelerated coking method to study the coking sites of γ-Al2O3 surface. The carbon species formed on γ-Al2O3 surface were analyzed by CAT-CS, TG-MS, IR-OH, and Py-IR techniques. The results of characterization by CAT-CS and TG-MS techniques indicated that the carbon species formed during the chemical static adsorption process is mainly composed of the reversibly adsorbed coke precursors with a lowly-condensed state, while that formed after the accelerated coking process is probably related with the irreversibly adsorbed coke deposits with a highly-condensed state. The results of characterization by IR-OH and Py-IR techniques further implied that the formation of the two kinds of carbon species, i.e., coke precursors and coke deposits, are closely related with the basic hydroxyl groups and the strong Lewis acid sites on γ-Al2O3 surface. The results lead to a deep insight into the coking mechanism on the alumina surface.

Key words: 1-methylnaphthalene; γ-Al2O3; coking sites; chemical static adsorption; accelerated coking

1 Introduction

Owing to the increasingly stringent environmental regulations, the production of ultra-low sulfur diesel fuel remains an essential problem for refiners. During the hydrodesulfurization (HDS) process, the dramatic increase in reaction severity (e.g., reaction temperature) and the substantial extension in reaction time all increase the deactivation rate and reduce the cycle lengths of hydrodesulfurization catalysts[1]. Besides, faced up to heavier and deteriorative crudes, refiners tend to blend a certain proportion of FCC light cycle oil (LCO) with straight-run gas oil (SRGO) to increase the utilization of crude oil and economic benefits. However, the addition of LCO to SRGO increases the contents of aromatic and nitrogen compounds and can substantially reduce the apparent reaction rates, which is not in favor of the longcycle operation of HDS catalysts[2-4]. Therefore, how to decrease the deactivation rate of HDS catalysts and prolong the operating cycle of refinery equipment become a key issue for the production of ultra-low sulfur diesel over a long period of time.

It is generally believed that two main causes are responsible for catalysts deactivation[2,5-6]. The first one is the accumulation of coke, which is prone to blocking a part of catalyst pores and preventing reactive molecules from gaining access to the active sites. Furthermore, the formation of coke on metallic sites can cover a part of active sites and induce the modification of active phase[7-8]. The second one is the modification of active phase, which means that a part of metal particles interacting weakly with the support can be liable to aggregation or segregation from active phase under severe reaction conditions[9-10]. Compared with the modification of active phase, coke deposition is generally believed to be the primary cause of catalysts deactivation[11-13]. It is well accepted that the coking process mainly contains two stages, that is, the reactants are f irst chemisorbed on the surface of the support to form the coke precursors, and then the coke precursors undergo a series of condensation reactions to form coke deposits. In order to f ind a solution to inhibit the formation of coke deposition and finally solve the problem of catalysts deactivation, it is essential to accurately define the coking sites on catalyst surface. So far, researchers have conducted extensive research on the coking sites. Absi-Halabi, et al.[14]identified that the initial deactivation process involved the deposition of a relatively large amount of coke resulted from adsorption of coke precursors on acidic sites of the support surface, and the use of less acidic support such as the carbon supported catalyst could effectively reduce the extent of coking. Furimsky, et al.[5]believed that both the Lewis and Br?nsted acid sites might take part, but could play different roles in the process of coking. The Lewis acid sites were strongly adsorbed with the coke precursors to prolong their residence time on the support and the Br?nsted acid sites provided the protons to form carbonium cations, which would finally form coke deposits after a series of polycondensation reactions. Scaroni, et al.[15-17]discovered that the prior adsorption of pyridine on the support decreased the concentration of Lewis acid sites and the amount of coke deposits, which suggested that the Lewis acid sites were involved in the coking process. Appleby, et al.[18]found that the coking sites were closely related with the Br?nsted acid sites, and when the basicity of the coke precursors or the concentration of Br?nsted acid sites increased, the coke deposits on the support surface also increased, which was consistent with the results obtained by Flockart et al.[19]

According to the above study, it is generally accepted that the coking sites should be mainly originated from the potential sites (Lewis and Br?nsted acid sites) on the support surface. However, questions still remain with regard to the coking mechanism on the Lewis and Br?nsted acid sites. In addition, these progresses are only related to the Lewis and Br?nsted acid sites without mentioning other interaction sites existing in the alumina surface, including the basic, neutral, and acidic hydroxyl groups, which can also play a significant role in the coking process. In this article, the typical bicyclic aromatic hydrocarbon (1-methylnaphthalene) in the FCC light cycle oil was selected as the research object, with which the support γ-Al2O3was treated by using the chemical static adsorption method and the accelerated coking method, respectively. Then the properties of carbon species and the coking sites were analyzed by CAT-CS, TG-MS, IR-OH, and Py-IR techniques.

2 Experimental

2.1 Samples preparation

The support used thereby was an industrial γ-Al2O3(denoted as Al) with a specific surface area of 285 m2/g and a pore volume of 0.79 mL/g. The pre-adsorption samples denoted as 1-MN/Al were prepared by using the chemical static adsorption method. Firstly, γ-Al2O3was treated with 1-methylnaphthalene (denoted as 1-MN) by the wet impregnation method. Then the prepared sample was left overnight and washed with toluene in a Soxhlet unit at 200 °C for 8 h and dried at 80 °C for 5 h. The precoked samples denoted as 1-MN/Al-AC were prepared in a continuous flow fixed-bed micro-reactor under the conditions covering a pressure of 4.0 MPa, a temperature of 400 °C, a catalyst loading of 0.50 g, a feed f low rate of 0.50 cm3/min using 40% of 1-MN inn-decane as the pre-coking feed, a H2flow rate of 100 cm3/min, and a precoking time of 5 hours. After the accelerated precoking treatment, the sample was washed with toluene in a Soxhlet unit at 200 °C for 8 h and dried at 80 °C for 5 h.

2.2 Samples characterization

The amount of carbon in the samples was measured by an EMIA-320-V carbon and sulfur analyzer (CAT-CS). The properties of coke species were characterized by a NETZSCH STA 409PC-QMS403C thermal analyzer and a mass spectrometer operating at an air f low rate of 100 mL/min (TG-MS). A linear heating rate of 10 °C/min was used for preheating the samples from 30 °C to 800 °C throughout the test. In the following discussion, the curve-fitting analysis of CO2formation profiles was completed using the manual f it with pure Gaussian type peaks. Because the formation profiles of H2O and CO2detected between 200 °C and 600 °C were originated from the coke combustion, the H/C ratio of coke could be calculated according to the integral area of the profiles of H2O andThe surface hydroxyl groups of alumina were determined on a Nicolet 870 Fourier-transform infrared spectrometer (IR-OH). The samples used for IR spectrometric analyses were finely ground, pressed into a self-supporting wafer and placed into an infrared cell with KBr windows. The degassification of sample was carried out at 500 °C and 10-3Pa for 2 h prior to the adsorption of pyridine (Py-IR). The IR spectra were recorded after subsequent evacuation at 200 °C and 350 °C for 0.5 h, respectively. The concentration of acid measured at 200 °C and 350 °C represented the concentration of total acid sites and strong acid sites, respectively. The concentration of weak acid sites was obtained by subtracting the concentration of strong acid sites from the concentration of total acid sites.

3 Results and Discussion

3.1 Carbon-sulfur elemental analysis

In order to investigate the amount of carbon formed on the surface of γ-Al2O3, the samples were subjected to CAT-CS analysis, with the results listed in Table 1. The results showed that the carbon content of 1-MN/Al was 0.90%, which indicated that after impregnation and toluene extraction, only a small part of 1-MN was chemisorbed on the surface of γ-Al2O3. When the carbon content of 1-MN/Al-AC was higher and reached 3.31%, it could indicate that after the accelerated coking process, a large part of coke species existed on the surface of γ-Al2O3.

Table 1 The carbon and sulfur contents of the samples

3.2 TG-MS analysis

In order to investigate the properties of coke species on the surface of γ-Al2O3, the samples were characterized by TG-MS analyses, with the formation profiles of H2O and CO2presented in Figure 1. Figure 1 shows that the formation profiles of H2O (Figure 1 (a)) of the samples were almost similar, whereas the CO2profiles of the two samples were obviously different and the peak intensity of 1-MN/Al-AC was much higher than that of 1-MN/Al (Figure 1 (b)), further indicating that the carbon amount of the carbon species deposited on the former was much more than the latter.

Figure 1 TG-MS results of the samples

To accurately analyze the relative peak positions of the formation profiles of H2O and CO2, the curve-fitting analysis was performed, with the deconvolution results shown in Figure 2 and Figure 3, respectively. Figure 2 shows that the H2O peaks of 1-MN/Al were located at 135 °C, 245 °C, 370 °C, and 450 °C and those of 1-MN/Al-AC were located at 145 °C, 250 °C, 390 °C, and 505 °C. Figure 3 shows that the CO2peaks of 1-MN/Al were located at 330 °C and 450 °C, whereas those of 1-MN/Al-AC were located at 340 °C, 440 °C, and 520 °C. Specifically, the H2O peaks at 135 °C or 145 °C were mainly assigned to the removal of physically adsorbed water from the surface of the samples, and the peaks at 245 °C or 250 °C might be assigned to the removal of chemically combined water and the combustion of a part of reversibly adsorbed coke precursors with a lowlycondensed state. Upon combining the deconvolution results of H2O and CO2, both the H2O peaks at 370 °C, 390 °C, and 450 °C and the CO2peaks at 330 °C, 340 °C, 440 °C, and 450 °C could be assigned to the combustion of reversibly adsorbed coke precursors with a lowlycondensed state, and the H2O peak at 505 °C and the CO2peak at 520 °C could be assigned to the combustion of irreversibly adsorbed coke deposits with a highlycondensed state[12,20]. Besides, according to the related literature[2,21], the combustion peak temperature of coke was an important factor used to measure the condensation degree of coke, and the higher the combustion peak temperature of coke, the higher the condensation degree of coke would be. Therefore, since the combustion peak temperature of coke of 1-MN/Al-AC was higher than that of 1-MN/Al, this fact indicated that the condensation degree of coke of the former was higher. Meanwhile, since the H/C atomic ratio of coke could directly ref lect the condensation degree of coke[8], the combination of the above-mentioned result with the analysis results (H/C(1-MN/Al)=14.4, H/C(1-MN/Al-AC)=2.88) could draw up the same conclusion that the carbon species on 1-MN/Al-AC was much more condensed than that on 1-MN/Al.

Based on the above analysis in combination with the results of CAT-CS analyses, it can be inferred that for 1-MN/Al, the carbon species chemisorbed on the surface of γ-Al2O3was mainly composed of the reversibly adsorbed coke precursors with a lowly-condensed state, while for 1-MN/Al-AC, the carbon species was mainly composed of the irreversibly adsorbed coke deposits with a highly-condensed state.

Figure 2 The deconvolution results of MS-H2O of the samples

Figure 3 The deconvolution results of MS-CO2 of the samples

3.3 IR-OH analysis

In order to study the influence of carbon species on the types and distribution of hydroxyl groups on the surface of the support, the samples were characterized by IR-OH, with the results shown in Figure 4. The results of Figure 4 indicate that the types of hydroxyl groups of 1-MN/Al and 1-MN/Al-AC were not notably changed in comparison with γ-Al2O3, and the peaks centered around 3 775 cm-1, 3 730 cm-1, and 3 674 cm-1were assigned to the basic, neutral, and acidic hydroxyl groups, respectively[22-24]. According to the relative peak positions of three types of hydroxyl groups, the curve-fitting analysis was performed with pure Gaussian type peaks to obtain the distribution of hydroxyl groups, with the deconvolution results shown in Figure 5. Meanwhile the concentration of each type of hydroxyl groups could be calculated by the integral area of fitted hydroxyl groups divided by the mass of the samples[25-26], with the detailed information of fitted hydroxyl groups listed in Table 2. According to the results of Table 2, the concentrations of the basic, neutral and acidic hydroxyl groups all decreased in the following order: Al＞1-MN/Al＞1-MN/Al-AC. Specifically, the concentration of basic hydroxyl groups of 1-MN/Al decreased much more than that of other types of hydroxyl groups compared with Al, indicating that the coke precursors were probably originated from the adsorption of 1-MN on the basic hydroxyl groups of γ-Al2O3surface. For 1-MN/Al-AC, the same change trend could be obtained as compared with 1-MN/Al, which indicated that the coke deposits with a highly-condensed state were also mainly generated from the interaction between the coke precursors with the basic hydroxyl groups on the surface of γ-Al2O3.

Figure 4 IR-OH results of the samples

Table 2 Detailed analysis data of fitted hydroxyl groups

3.4 Py-IR analysis

Figure 5 The deconvolution results of IR-OH of the samples

In order to study the influence of carbon species on the acid sites and acid type distribution on the surface of the support, the samples were characterized by Py-IR, with the results shown in Figure 6. According to the results of Figure 6, for the sample of Al, 1-MN/Al and 1-MN/Al-AC, the absence of a peak at around 1 540 cm-1indicated that there were no Br?nsted acid sites on the surface of the samples before and after carbon species were formed on the surface of γ-Al2O3. The presence of a peak at around 1 450 cm-1demonstrated that the Lewis acid sites were present on the surface of the samples. To accurately analyze the distribution of Lewis acid sites with different acid strengths, the characteristic peaks at around 1 450 cm-1of the three samples after evacuation at 200 °C and 350 °C were integrated, respectively, and the results were shown in Table 3. Judging from the data of Table 3, it can be seen that the concentration of the total, strong, and weak Lewis acid sites all decreased in the following order: Al＞1-MN/Al＞1-MN/Al-AC. Specifically, the concentration of the strong Lewis acid sites of 1-MN/Al decreased much more than that of the weak Lewis acid sites compared with Al, and the same change trend could be obtained by 1-MN/Al-AC in comparison with 1-MN/Al, indicating that the strong Lewis acid sites were the major interaction sites on the surface of γ-Al2O3either in the chemical adsorption or condensation process of carbon species.

Figure 6 Py-IR results of the samples

Table 3 Concentration of different acid sites of the samples

According to the relevant literature reports[22-23], the Lewis acid sites and hydroxyl groups were adjacent to each other on the surface of the support, and the weak Lewis acid sites had neutral hydroxyl groups nearby, the medium strong Lewis acid sites had acidic hydroxyl groups nearby, while the strong Lewis acid sites had basic hydroxyl groups next to them. During the process of Py-IR analysis, the pyridine molecules adsorbed on the Lewis acid sites could simultaneously achieve hydrogen-bonding with their adjacent hydroxyl groups, from which it could be inferred that the interaction of hydroxyl groups with the carbon species formed in the adsorption or condensation process, respectively, was not determined by the hydroxyl groups themselves but by the types of the Lewis acid sites. Based on the above discussion, in combination with the analysis results of Py-IR confirming that the carbon species tended to interact with the strong Lewis acid sites, it could be reasoned that the carbon species tended to interact with the basic hydroxyl groups adjacent to the strong Lewis acid sites, which was consistent with the analysis results of IR-OH. Owing to the fact that the analysis results of IR-OH and Py-IR could be mutually verified, the conclusions regarding the coking sites of the carbon species on the surface of γ-Al2O3were reliable.

4 Conclusions

1) The support γ-Al2O3was treated with 1-MN by using the chemical static adsorption method and the accelerated coking method, respectively. The results of characterization by CAT-CS and TG-MS techniques indicated that, for 1-MN/Al, the carbon species was mainly composed of the reversibly adsorbed coke precursors with a lowlycondensed state, while for 1-MN/Al-AC, the carbon species was mainly composed of the irreversibly adsorbed coke deposits with a highly-condensed state.

2) The results of characterization by IR-OH and Py-IR techniques indicated that, both the coke precursors formed by the adsorption of reactant molecules and the coke deposits formed by the condensation of coke precursors are closely related with the basic hydroxyl groups and the strong Lewis acid sites on the surface of γ-Al2O3.

Acknowledgement:The authors gratefully acknowledge the support from the National Key Basic Research Program of China (Grant 2017YFB0306603).