Preparation of 2-Methylnaphthalene from 1-Methylnaphthalene via Catalytic Isomerization and Crystallization

Large amounts of residual 1-methylnaphthalene are generated when 2-methylnaphthalene is extracted from alkyl naphthalene. In order to transform waste into assets, this study proposes a feasible process for preparing 2-methylnaphthalene from 1-methylnaphthalene through isomerization and crystallization. The 1-methylnaphthalene isomerization was carried out in a fixed-bed reactor over mixed acids-treated HBEA zeolite. The results showed that acidic properties of catalysts and reaction temperature were associated with the 2-methylnaphthalene selectivity, yield and catalytic stability. At a high reaction temperature of 623 K, the 2-methylnaphthalene yield was 65.84 %, and the deactivation rate was much lower. The separation of reaction products was then investigated by two consecutive crystallization processes. Under optimal conditions, the 2-methylnaphthalene purity attained 96.67 % in the product, while the yield was 87.48 % in the refining process. Copyright © 2018 BCREC Group. All


Introduction
The production of C10 aromatics are the residual oils after trimethylbenzene and indane are extracted from C9 aromatics, and have reached more than 300,000 tons per year in China. Using C10 aromatics as solvents or fuel oil is both inefficient and polluting, because they are made up of about 30 % alkyl naphthalenes. Among alkyl naphthalenes, 2-methylnaphthalene (2-MN) is a well-known useful intermediate for producing polyethylene naphthalene (PEN) and vitamin K [1][2][3]. In our previous study [4], 2-MN was attained from alkyl naphthalenes via side-stream distillation and continuous crystallization processes. However, the residual component was mainly 1-methyl-naphthalene (1-MN), the industrial demand for which is much lower than that for 2-MN. If we could obtain 2-MN from 1-MN, residual C10 aromatics could be maximally utilized with less waste.
Past studies [5][6][7][8][9][10][11] have demonstrated HBEA zeolite to be an effective catalyst for 1-MN isomerization owing to its thermal stability, appropriate pore size distribution and high activity in comparison with those of other zeolites. However, due to its initial defective structure, detrimental by-products are unavoidable during the synthesis process, and its catalytic life of isomerization is short. Modifications of zeolites, such as: silylation [12], stream [13], and dealumination [14][15][16][17], are advantageous for improving the stability and selectivity of catalytic reaction. Our previous report has studied the HBEA zeolite modified by organic acid [11], and the modified catalyst demonstrated high selectivity for the conversion of alkyl naphthalene. Dealumination is accomplished by hydrolysis of Al-O-Si bonds with inorganic and/or organic acids. It is obvious that the acids used in the treatment are responsible for the acidic properties and catalytic performance of HBEA zeolite. Nevertheless, the mixture of inorganic and organic acids utilized for the dealumination of HBEA and the catalytic deactivation rate of mixed acids-treated zeolite for 1-MN isomerization have scarcely been investigated.
The product of 1-MN isomerization contains both 1-MN and 2-MN due to the thermodynamic equilibrium of isomerization reaction. The components of this mixture have a large difference in melting point, but a small difference in volatility. As reported in our previous study [4], crystallization is an effective method for separating the isomers mixture.
In order to make full use of the 1-MN resulting from residual C10 aromatics, this work carried out the 1-MN isomerization over modified HBEA zeolite. The modified zeolite was characterized by various methods, and its catalytic properties were investigated in a fixed-bed reactor. Furthermore, in order to obtain 2-MN with high purity and yield, two-fold crystallization for the separation of methylnaphthalene isomers was also studied.

Source of Chemicals
The 1-MN (> 98 % purity) was purchased from Aladdin Industrial Inc., Shanghai, China. Hydrochloric acid and oxalic acid were both analytical regents obtained from Sinopharm Chemical Regent Co., Ltd., Shanghai, China. Parent HBEA zeolite was synthesized by the method described in Matsukata et al. [18].

Catalyst Preparation
The parent HBEA zeolite was activated at 823 K (2 K/min heating rate) for 5 h before the modification. The dealuminated catalyst was prepared by impregnating HBEA zeolite into 0.1 mol/L solution of mixed hydrochloric and oxalic acids (hydrochloric/oxalic = 1) at 343 K for 1 h. After that, the sample was filtered, washed, dried and calcined at 823 K (2 K/min heating rate) for 4 h. The synthesized catalyst was denoted as Mix-HBEA.

Characterization
X-ray diffraction (XRD) was recorded on a D/max-2500 (Rigaku, Japan) using Cu-Kα radiation and operating at 100 mA and 40 kV. Fourier transform infrared (FTIR) spectroscopy was conducted on a Tensor 27 (Bruker, Germany) spectrometer in the mid-IR region (400-4000 cm -1 ). Surface acidity was performed with FTIR spectrometry after adsorption of pyridine (Py-FTIR), using a Vertex 70 (Bruker, Germany) spectrometer coupled with a conventional high vacuum system. The molar extinction coefficients for pyridine on the Brönsted (1545 cm -1 band, εB) and Lewis acid sites (1454 cm -1 band, εL) were identified as 1.23 cm/μmol and 1.73 cm/μmol, respectively [19]. NH3 temperature-programmed desorption (NH3-TPD) was carried out on a TP-5080 (Xianquan, China) instrument equipped with a thermal conductivity detector. Thermogravimetric (TG) analysis was measured using a Diamond TG/DTA (Perkin Elmer, USA) instrument in the presence of oxygen.

Reaction and Separation
The isomerization of 1-MN was performed in a fixed-bed reactor. After 1.6 g of catalyst was loaded in the center of the reactor, the isomerization was carried out in the presence of nitrogen under atmospheric pressure. The flow-rate of 1-MN and N2 were 1.3 h -1 weight hourly space velocity (WHSV) and 20 mL/min, respectively. Products were sampled and analyzed at regular intervals. The 1-MN conversion was based on the 1-MN concentration change, and 2-MN selectivity was defined as the mass ratio of 2-MN to 1-MN conversion.
A home-made apparatus [20] was then utilized to crystallize and filter the products of the isomerization reaction. This apparatus consists of a jacketed crystallizer, a homothermal filter and a filtrate collector. The temperature of the apparatus was controlled by a circulating pump, using cold glycol as a cooling medium. The 2-MN yield in the crystallization process was calculated as follows: (1) where Y2-MN is the 2-MN yield in the reaction process, Ycrystal is the crystal yield of the crystallization process, and C2-MN is the 2-MN purity of the crystal in the crystallization process.

Characterization of Catalyst
The crystal structure and lattice vibration of Mix-HBEA were measured by XRD and FTIR, respectively. The XRD results plotted in Figure 1a show typical reflections of the zeolite HBEA at 7.8° and 22.5°, with no other significant peaks. As shown in Figure 1b, the bands at 1090 and 795 cm -1 are associated with the symmetric and asymmetric stretching vibrations of the TO4 units of zeolite, respectively. At 569 cm -1 , the adsorption band represents a characteristic of the BEA framework structure in the Mix-HBEA catalyst [21]. These results suggest that treatment with mixed acids preserves the zeolite structure of HBEA.
NH3-TPD analysis was carried out to estimated the types and number of acid sites on the surface. The results are depicted in Figure  2a. NH3 desorption peaks occurred in the ranges of 100-300 °C and 300-700 °C are as- cribed to weakly acidic sites and strongly acidic sites, respectively [22,23]. The amounts of weak and strong acid were calculated based on the areas of the two peaks, and the results are listed in Table 1. The amount of strong acid is clearly higher than that of weak acid, and these weak acid sites are unable to activate the isomerization reaction [24].
Further information on the strong Brönsted and Lewis acidic properties of Mix-HBEA were then determined through Py-FTIR, and the results are illustrated in Figure 2b. The bands at 1454 and 1545 cm -1 are related to pyridine adsorbed on the Lewis acid sites and Brönsted acid sites, respectively [25]. The band at 1490 cm -1 is assigned to pyridine adsorbed on both Brönsted and Lewis acid sites. Therefore, the amounts of Lewis and Brönsted acid sites can be measured according to the absorption intensities of the bands at 1454 and 1545 cm -1 , respectively. As listed in Table 1, the ratio of strong Brönsted to Lewis acid sites was calculated based on Figure 2b [26]. Clearly, strong Brönsted acid sites far outnumber the Lewis acid sites for Mix-HBEA.

Catalytic Performance and Coke Formation of Catalyst
In order to investigate the conversion of 1-MN to 2-MN, the isomerization of 1-MN was performed over the Mix-HBEA catalyst at 573 K and 623 K. As demonstrated by past studies [8,27], the 1-MN conversion and 2-MN selectivity are decreased distinctly at temperature lower than 573 K and higher than 623 K, respectively. The other reaction conditions were as follows: WHSV was 1.3 h -1 , the time on stream (TOS) was 1 h and the carrier gas flowrate was 20 mL/min. Table 2 lists the conversion results over the Mix-HBEA catalyst. Like in our previous study [11], disproportionation of 1-MN to NA and DMN is observed in conjunction with the isomerization of 1-MN to 2-MN.
At 573 K, the selectivity and yield of 2-MN over Mix-HBEA are 96.73 % and 67.10 %, respectively. These results suggest restrained disproportionation, as the 2-MN yield almost reaches the thermodynamic equilibrium of MN isomers. The results of NH3-TPD and Py-FTIR indicate that both the weak and strong Lewis acid sites for Mix-HBEA are fewer in number than the strong Brönsted acid sites. These results imply that the Lewis and strong Brönsted acid sites in the catalyst act as active sites in the disproportionation [28,29] and isomerization [24,30] of 1-MN, respectively. When the reaction temperature increases from 573 to 623 K, the conversion of 1-MN is increased to 71.98 %, while the yield of disproportionation products is slightly enhanced [8]. As a result, the 2-MN selectivity decreases to 91.46 %, while the yield of 2-MN remains at a high level.
The stability of the Mix-HBEA catalyst with TOS was also tested at 573 K and 623 K. As shown in Figure 3a,     ln y t -ln y 0 = -k d t (2) where, kd is the deactivation rate, t is the TOS, and y0 and yt represent the conversion on the fresh and fouled catalyst, respectively [27]. The change in ln y t with TOS is plotted in Figure  3b, and the values of kd were calculated from the linear fittings. As listed in Table 2, the deactivation rate of Mix-HBEA at 623 K is less than one fifth that at 573 K, indicating that the deactivation rate decreases with increasing reaction temperature [31]. Thus, the isomeriza-tion stability of Mix-HBEA is improved significantly when the reaction temperature increases from 573 to 623 K. Considering the 1-MN conversion, 2-MN yield and isomerization stability, 623 K appears to be an appropriate reaction temperature for 1-MN isomerization over Mix-HBEA. The slightly low reaction selectively is not an obstacle, as the byproducts are eliminated in the crystallization process that follows.
The formation of coke deposits on the used catalysts was determined by TG under temperature programmed oxidation, and the results are shown in Figure 4. In the 40-200 °C range, the weight loss is associated with adsorbed moisture and gas in the catalysts. Above 200 °C, the steady weight of the fresh catalyst indicates none of carbonaceous deposits in unused Mix-HBEA. Nevertheless, in the 200-700 °C range, the weight of the used Mix-HBEA de-  creases due to the oxidation of carbonaceous deposition in catalyst [26]. These coke deposited on the used catalyst account for the decrease of 1-MN conversion over Mix-HBEA after 10 h TOS at 573 K.

Crystallization
After the catalytic isomerization of 1-MN over the Mix-HBEA catalyst at 623 K, crystallization was carried out to refine 2-MN from the reaction product (see Table 2). As the quality of crystal products is remarkably influenced by different cooling conditions [32], the main factors including crystallization temperature, crystallization time and filtration time were examined in the crystallization process.
The effects of crystallization temperature on the separation process are plotted in Figure 5, where crystallization and filtration times were 180 and 20 min, respectively. The crystal yield decreases as the crystallization temperature increases. Meanwhile, as the crystallization temperature increases from 255 to 261 K, the 2-MN purity increases from 93.61 to 95.68 %, and the 2-MN yield also increases slightly. However, both the purity and yield of 2-MN decrease as the crystallization temperature increases further. These results suggest that crystallization at lower temperatures favors the nucleation and growth of 2-MN crystals, whereas the impurity would be incorporated into the growing crystals due to the burst nucleation at excessively low temperature [32,33]. Accordingly, a crystallization temperature of 261 K is optimal. Figure 6 illustrates the effects of crystallization time on 2-MN purity and the yields of 2-MN and crystal, where the crystallization temperature was 261 K and the filtration time was 20 min. As crystallization time increases from 2 to 5 h, the 2-MN purity and the yields of 2-MN and crystal increase distinctly. However, the purity and yield of 2-MN decrease simultaneously as the crystallization time increases further, despite a slight increase in crystal yield. It is indicated that the impurity can grow into the crystal lattice easily because of the similarity of their structure with 2-MN molecule [34]. Therefore, 5 h represents an appropriate crystallization time in the crystallization process. Figure 7 illustrates the effects of filtration time on 2-MN purity and yields of 2-MN and crystal, where the crystallization temperature was 261 K and the crystallization time was 5 h. The crystal yield decreases as the filtration time increases. Furthermore, as the filtration time increases from 5 to 20 min, 2-MN purity increases from 89.69 to 96.84 %, and 2-MN yield increases from 58.44 to 62.23 %. When the filtration time increases further from 20 to 30 min, both 2-MN purity and yield decrease. One possible reason for this phenomenon is that several crystals might be dissolved and filtrated in the impurity [35]. Thus, an advisable filtration time is 20 min.
The above results indicate that optimal parameters for the crystallization process are as follows: a crystallization temperature of 261 K, a crystallization time of 5 h, and a filtration time of 20 min. Under these conditions, the 2-MN purity and yield are 96.84 % and 62.23 %, respectively, and the crystal yield is 41.77 %. Moreover, the attaining filtrate of the crystallization process was crystallized again under the same conditions, after which the 2-MN pu- rity and yield are 96.25 % and 25.25 %, respectively, and the crystal yield is 17.05 %. Overall, after two consecutive crystallization processes, the 2-MN purity is 96.67 % and the total 2-MN yield of crystallization is 87.48 %. Thus, it is feasible to apply isomerization and crystallization to prepare highly purified 2-MN from 1-MN.

Conclusion
In this process, the isomerization of 1-MN over the Mix-HBEA catalyst and two-fold crystallizations to refine 2-MN were investigated. The Mix-HBEA catalyst presented fewer weak acid sites and strong Lewis acid sites than strong Brönsted acid sites, and it preserved the HBEA zeolite structure. At 623 K, the 2-MN yield was 65.84 % and the deactivation rate was 0.003 h -1 during the 1-MN isomerization process. After obtaining these results, the parameters for crystallization temperature, crystallization time and filtration time during the separation procedure were studied. Under optimal conditions, 2-MN attained 96.67 % purity, and the 2-MN yield reached 87.48 % after two consecutive crystallizations. Overall, the contribution of this paper lies in outlining a process by which residual 1-MN from C10 aromatics can be transformed to the more valuable 2-MN.