Molded MFI nanocrystals as a highly active catalyst in a methanol-to-aromatics process

Abstract

ZSM-5 nanocrystals with a short b-axis length exhibited higher activity and lower coke selectivity in a methanol-to-aromatics reaction, compared to microcrystals, whether in the pristine state or in molded particles with an average size of 100 microns by adding silica-sol and kaolin clay required by industrial scale application.

---

ZSM-5 zeolite, with a high surface area, tunable porosity and tunable acidity, has been widely used as the catalyst in transforming methanol into hydrocarbons (olefins,^1,2 gasoline^3,4 or aromatics^5–7). The latter methanol-to-aromatics reaction (MTA) was carried out at high temperatures up to 450–500 °C, resulting in serious pore blockage by coke deposition⁸ and the rapid deactivation of the zeolite-based catalyst.^9,10 Decreasing the size of zeolite to the nanoscale or creating mesopores within ZSM-5 structures would shorten the diffusion path of large molecules in channel of zeolite in b-axis direction so as to enhance the activity of the catalyst.^11,12 As a result, the stable life time was increased from several hours with microsized catalyst^13,14 to several ten to hundreds hours as using nanosized catalyst.^15,16 However, it was still not long enough to ensure a constant operation in industrial scale operation as using fixed or packed bed reactor. In this case, a continuous reaction-catalyst regeneration in fluidized bed system, mature used in oil refinery or methanol-to-olefins, would be crucial to the scale up of MTA process. It called for the molding of the nanosized zeolite to the powder with average size around 100 micrometeres. But there is rare report on their molding effect on their catalytic performance of nanosized zeolite, probably due to the limited amount of nanosized zeolite.

In the present work, we reported the comparative study of the catalytic performance of coffin-shaped MFI nanocrystals or the MFI microcrystals with or without molding. Detailed information of BET surface area, pore size distribution, pore volume of different samples was investigated as well as the MTA performance. It was found that Zn doped pristine coffin-shaped ZSM-5 nanocrystals (denoted as Zn/CZ5) with short b axis length exhibited far higher yield of aromatics, compared to that with pristine ZSM-5 microcrystals (assigned as Zn/MZ5). Molding Zn/CZ5 and Zn/MZ5 with 2–3 microns into powders with 100 μm size (denoted as Zn/CZ5-M and Zn/MZ5-M, respectively), both resulted in the significant decrease of their activities or performances. The aromatic product distributions were changed significantly, where the formation of trimethylbenzene was suppressed since the loss of huge external surface of nanosized zeolite after molding with silica gel. However, the molded nanocrystals (Zn/CZ5-M) still exhibit higher performance than that molded microcyrstals (Zn/MZ5-M), or even comparable activity to that of pristine microcrystals (Zn/MZ5). These comparisons indicated that the nanocrystals of zeolite had important potential application toward scale up.

Experimentally, we used urea as additive in the preparation of MFI crystal by hydrothermal method. In detail, 13.1 g TPAOH, 11.2 g TEOS, 2.0 g urea, 0.3 g Al(NO₃)₃·9H₂O, 0.1 g NaOH and 0.1 g IPA were added into 18.4 g H₂O with stirring for 1–2 h. The solution was transferred into an autoclave and was heated from room temperature to 180 °C with a rate of 15 °C h⁻¹, and then kept at 180 °C for 48 h. The obtained Na–ZSM-5 crystals after filtration were washed by deionized water for 3 times, subsequently dried at 90 °C and calcined at 550 °C in air for 5 h. Then Na–ZSM-5 was converted into NH₄–ZSM-5 by cation-exchange in a NH₄NO₃ solution for 3 times. H-type ZSM-5 was obtained by calcining the NH₄–ZSM-5 powder at 550 °C for 5 h. In addition, we purchased the ZSM-5 microcrystals with similar Si/Al ratio (or similar acidity) for comparison. For molding of the zeolite, zeolite was mixed with silica sol (as binder) and kaolin clay (as mechanical additive) in a solution with solid concentration of 15%. Weight ratio of zeolite, whether is nano or micro, were controlled to 30% for fair comparison. The solution was quickly sprayed into a granulator with temperature of 300–350 °C. Water was quickly heated into vapor and the solids condensed to become powder. To test the catalytic performance of MFI in model reaction of methanol to aromatics, 3 wt% Zn was doped based on the weight of ZSM-5 by an incipient wetness impregnation method using aqueous solutions of Zn(NO₃)₂·6H₂O. After drying and calcinating in air at 550 °C for 5 h. The reaction was performed at 475 °C and normal pressure with a space velocity of methanol of 0.8 h⁻¹.

The as prepared Zn/CZ5 was mostly coffin shaped sample with length of 200–300 nm (c axis direction) and thickness of 50 nm (b-axis direction) (Fig. 1a). The size or morphology of zeolite was quite uniform. Doping with zinc did not change their morphology. And the lattice can be clearly observed, since species of zinc oxides distributed uniformly in the matrix of zeolite (Fig. 1c). In comparison, microsized Zn/MZ5 was a product mixture of 600 nm to 2 microns. The length in a, b, c-axis direction was quite similar. Large size resulted in the black zone in TEM image (Fig. 1d), which was unable to be penetrated by electron beam. Similar, the lattice near the edge was clearly seen, due to the uniform distribution of zinc species inside the matrix. The formation of such perfect MFI nanocrystals of Zn/CZ5 was attributed to the addition of urea and the increased addition of NaOH. Urea was effective to hinder the growth along the b-axis, due to the formation of inert bonds with Si–O–Al groups.¹⁷ Transition of MFI structure from that with similar a-, b-axis length to that with short b-axis length was observed. The transition, actually, decreased the edge number of MFI crystals. And similar clean (010) and (100) surface was observed.¹⁷

Fig. 1 (a) SEM image of Zn/CZ5, (b) SEM image of Zn/MZ5, (c) TEM images of Zn/CZ5, (d) TEM image of Zn/MZ5.

After molding with silica sol and kaolin clay, both the ordered structure of nano or micro zeolite disappeared. The morphology of powders was similar to be spherical particles (Fig. 2a). Particle size distribution was centered in 100 micrometers and the largest particle is smaller than 250 microns. Only 10% particles smaller than 15 microns, in agreement with standard quality control for a spray drying method. The smallest particles were still larger than 3 microns, indicating nanocrystals were fully capsulated by the silica layer. Ar adsorption suggested that pristine MFI nanocrystals and microcrystals were both micropore-dominant zeolite. There was no obvious Ar intake in high P/P₀ region (Fig. 2c). Surface area of pristine MFI nanocrystal was 427 m² g⁻¹ in total, among of which 280 m² g⁻¹ belonged to the external surface. Total pore volume of them was 0.54 mL g⁻¹, which included 0.20 mL g⁻¹ and 0.32 mL g⁻¹ for micropores and stacking pores, respectively. The Si/Al ratio determined by ICP was 64.5. In comparison, surface area of Zn/MZ-5 is 383.2 m² g⁻¹ and total pore volume was 0.32 mL g⁻¹, among of which 0.23 mL g⁻¹ belonged to micropores, and 0.09 mL g⁻¹ belonged to mesopores. In obvious contrast, the molding resulted in the decrease of surface area to 143.5 and 137 m² g⁻¹ for nano and microsamples. The surface area gap between Zn/CZ5 and Zn/MZ5 was filled. But the total volume of Zn/CZ5-M is 0.32 mL g⁻¹, 2 times that of Zn/MZ5-M. Detailed pore size distribution suggested that Zn/MZ5 exhibited the highest pore volume below 1 nm (Fig. 2d). Zn/CZ5 also had sufficient micropores smaller than 1 nm. Mesopores of Zn/CZ5 and Zn/MZ5 were only centered in 2–3 nm. Molding of them resulted in both the significant decrease of micropores by 60–80% and the disappearance of 2–3 nm mesopores, but significant increase of the mesopores at 10–30 nm. Specifically, mesopres of 30–100 nm of the nanozeolite was significantly suppressed, validating the stacking pores of nanozeolite was fully filled by other media. Probably because the microcrystals were agglomerate themselves, its mesopore volume at 30–100 nm was quite small. But molding of them further made these pores completely disappeared. These results confirmed that molding produced large particles with decreased surface area and decreased pore volumes. From NH₃-TPD curves (Fig. S1†), it can be seen that the amount and strength of acid sites over these two catalysts were similar. From the Py-FTIR spectra (Fig. S2†), the peaks at 1445 and 1540 cm⁻¹ were representative of Lewis and Brønsted acid sites, respectively, and the peak at 1490 cm⁻¹ represented both Lewis and Brønsted acid sites.¹² It was also noteworthy that the amount of L acid and B acid were similar between the two samples. There were minor acidic differences between the Zn/CZ5-M and Zn/MZ5-M, indicating that the effect of acidity on the catalytic performance can be ignored between these two samples.

Fig. 2 (a) SEM image of Zn/CZ5-M; (b) particle size distribution of Zn/CZ5-M; (c) comparison of Ar adsorption/desorption isotherm and (d) pore size distribution of Zn/CZ5, Zn/CZ5-M, Zn/MZ5 and Zn/MZ5-M.

As follows, the catalytic performance of Zn/CZ5 and Zn/MZ5 were firstly compared in MTA reaction. Yields of aromatics was 98% over Zn/CZ5, but was only 60% for Zn/MZ5. It was well known that nanozeolite with short b axis length exhibited excellent catalytic performance, compared to that based on microcrystals.¹⁸ Actually, when aromatics became the majority product, the diffusion inside channel or pore mouth became a molecule size-dependent process. For instance, smaller para-xylene exhibited a diffusion rate nearly 1000 times that of larger meta-xylene. In this case, pore diffusion capability of different catalysts would be revealed clearly. Also the excellent performance was due to the doping ZSM-5 with Zn, which formed the Lewis acids in large amount to exhibit much higher dehydrogenation ability. As to the product distribution, yield of C₉₊ with Zn/SZ5 was larger than 75%, due to the excess alkylation of xylenes on external surface,^19,20 considering C₉ (trimethylbenzene, about 0.61 nm in diameter) was unable to be produced in large amounts inside the channel. It directly resulted in the low yield of xylenes. In comparison, xylenes were the dominant components in product over Zn/MZ5 and the yield was even higher than the total yield of benzene, toluene and C₉₊. But the significant difference indicated that the low efficiency of microcrystals, due to the poor diffusion capability, resulted in the formation of C₁–C₅ hydrocarbons in large amount. It was quite reasonable considering the aromatization was a slow reaction and there existed dual aromatic-cycle and olefin-cycle inside the ZSM-5 channel,²¹ according to the hydrocarbon pool mechanism. Similar, the molding of nanocrystals or microcrystals into much larger powders weaken the diffusion ability significantly (Fig. 3). In detail, the aromatics yield dropped from 98% that Zn/CZ5 to 54% that molded Zn/CZ5-M. And the aromatics yield dropped from 60% that Zn/MZ5 to 39% that molded Zn/MZ5-M. Although the aromatization ability of nanocrystals dropped by 44%, larger than the drop (21%) as using microcrystals after molding, the absolute value of aromatics yield (54%) was only 6% less than the pristine microcrystals. In addition, the small particle size of pristine microcrystals was inappropriate for the fluidization operation in scale up and the aromatics yield (39%) for molded microcrystals is unacceptable for the scale up. As to the product distribution, C₉₊ hydrocarbons dropped significantly and their absolute value was close to that of xylenes with Zn/CZ5-M. Meanwhile, the yield of benzene and toluene increased drastically. In previous works, we validated that the formation of C₉₊ was mainly due to the external alkylation of xylenes and the formation of benzene and toluene was due to the dealkyaltion of xylenes.¹⁰ In this case, the molding resulted in the disappearance of huge pristine external surface of Zn/CZ5. In addition, although the external surface of microcrystals Zn/MZ5 was not so large, the molding also resulted in the loss of external surface. In this case, aromatics yield dropped by 21%. Similarly, coating Zn/ZSM-5 microcrystals with 8 nm SiO₂ layer exhibited similar pore-mouth blockage effect, resulting in the decreased yield of aromatics by 5–10%.¹⁰

Fig. 3 Product distributions of (a) Zn/CZ5 and Zn/MZ5, (b) Zn/CZ5-M and Zn/MZ5-M as a function of time-on-stream in methanol to aromatic reaction.

Generally, the large molded particles as the agglomerate of many core–shell structures, which had zeolite core and inert shell. In this case, the comparison above suggested that the decrease of aromatics yield of zeolite was the outer shell thickness dependent. Thick shell would exhibit larger resistance for the diffusion of aromatic species in the channel of zeolite and out of the channel of zeolite but still inside the molded particles. In order to quantitative evaluate the molding on the coke deposition of different catalyst, we carried out the TGA analysis (Fig. 4a). Coke selectivity was 2.6% for pristine Zn/MZ5 and was increased to 2.9% after molding the microsized zeolite. These data were comparable to the value (4–5%)²² of the molded catalyst in methanol-to-olefins process and that (6–7%)²³ of the molded catalyst in oil refinery process. In sharp contrast, coke selectivity was only 0.1% with pristine Zn/CZ5 (nanosized zeolite). Even after molding, the coke selectivity was increased to 0.3%, only about 1/9–1/10 that using microsized zeolite. These data, for the first time, validated that the diffusion resistance of large molecules cross the shell outside the zeolite is smaller than that inside the channel of zeolite. Structurally, nano zeolites possess higher volume and higher density inside the molded particles than the microsized zeolites, based on the same weight ratio in same agglomerate size after molding. In this case, the average diffusion length of large molecules from the external surface of the nano zeolite to the external surface of molded particles will be shorter compared to that of microsized zeolite. We presented an illustration to stress their structure difference (Fig. 4b). In this case, we validated the excellent diffusion capability and catalytic performance of the present MFI nanocrystals with short b length was the key for the acceptable performance after molding for scale up. On the other hand, we also stressed that the present work just compare the performance of zeolite with Si/Al ratio around 60. MTA called for a catalyst with much low Si/Al ratio and much stronger acidity and acids in larger amount.^24,25 The preparation of such kind of coffin-shaped nanocrystals and the related catalytic performance comparison needs the further investigation.

Fig. 4 (a) Thermal gravimeter (TG) & coke selectivity of Zn/CZ5, Zn/MZ5, Zn/CZ5-M and Zn/MZ5-M. Oxidation gas: 20% O₂/N₂, heating rate 10 °C min⁻¹; (b) illustration of molded microcrystals and molded nanocrystals.

In summary, we compared the performance of MFI nanocrystal and microcrystals with and without molding. The molding process de-creased the surface area, blocked the micropores of the zeolite, gave a significantly decreased catalytic performance in MTA process, changed the product distribution of aromatics and increased the coke selectivity accordingly. Although these trends were expected qualitatively, the use of MFI nanocrystals with short b-axis length still ensure their acceptable catalytic performance after molding, which was only 6% lower than that of the pristine microcrystals, but 15% higher than the molded microcrystals. It highlighted the importance of nanocrystals for potential scale up application in a fluidization system.

Acknowledgements

Y. Ma and N. Wang contributed equally to this work. The authors thank the support of NSFC program (21376135, 91434122 and 51236004) and CNPC Innovation Foundation of 2014D-5006-0506.

Notes and references

1. W. Wang, Y. Jiang and M. Hunger, Catal. Today, 2006, 113, 102–114.
2. W. Song, J. F. Haw, J. B. Nicholas and C. S. Heneghan, J. Am. Chem. Soc., 2000, 122, 10726–10727.
3. B. L. Bleken, D. S. Wragg, B. Arstad, A. E. Gunnæs, J. Mouzon, S. Helveg, L. F. Lundegaard, P. Beato, S. Bordiga, U. Olsbye, S. Svelle and K. P. Lillerud, Top. Catal., 2013, 56, 558–566.
4. F. Meng, Y. Wang and S. Wang, RSC Adv., 2016, 6, 58586–58593.
5. N. Wang, W. Qian, K. Shen, C. Su and F. Wei, Chem. Commun., 2016, 52, 2011–2014.
6. M. Conte, J. A. Lopez-Sanchez, Q. He, D. J. Morgan, Y. Ryabenkova, J. K. Bartley, A. F. Carley, S. H. Taylor, C. J. Kiely and K. Khalid, Catal. Sci. Technol., 2012, 2, 105–112.
7. Y. Ono, H. Adachi and Y. Senoda, J. Chem. Soc., Faraday Trans. 1, 1988, 84, 1091–1099.
8. H. Schulz, Catal. Today, 2010, 154, 183–194.
9. G. Q. Zhang, T. Bai, T. F. Chen, W. T. Fan and X. Zhang, Ind. Eng. Chem. Res., 2014, 53, 14932–14940.
10. J. G. Zhang, W. Z. Qian, C. Y. Kong and F. Wei, ACS Catal., 2015, 5, 2982–2988.
11. M. Choi, K. Na, J. Kim, Y. Sakamoto, O. Terasaki and R. Ryoo, Nature, 2009, 461, 246–249.
12. K. Shen, W. Z. Qian, N. Wang, C. Su and F. Wei, J. Mater. Chem. A, 2014, 2, 19797–19808.
13. M. Conte, J. A. Lopez-Sanchez, Q. He, D. J. Morgan, Y. Ryabenkova, J. K. Bartley, A. F. Carley, S. H. Taylor, C. J. Kiely, K. Khalid and G. J. Hutchings, Catal. Sci. Technol., 2012, 2, 105–112.
14. Y. Ono, H. Adachi and Y. Senoda, J. Chem. Soc., Faraday Trans. 1, 1988, 84, 1091–1099.
15. K. Shen, N. Wang, W. Z. Qian, Y. Cui and F. Wei, Catal. Sci. Technol., 2014, 4, 3840–3844.
16. K. Shen, W. Z. Qian, N. Wang, C. Su and F. Wei, J. Am. Chem. Soc., 2013, 135, 15322–15325.
17. Y. Liu, X. Z. Zhou, X. M. Pang, Y. Y. Jin, X. J. Meng, X. M. Zheng, X. H. Gao and F. S. Xiao, ChemCatChem, 2013, 5, 1517–1523.
18. Y. M. Ni, A. M. Sun, X. L. Wu, G. L. Hai, J. L. Hu, T. Li and G. X. Li, Microporous Mesoporous Mater., 2011, 143, 435–442.
19. S. Namba, A. Inaka and T. Yashima, Zeolites, 1986, 6, 107–110.
20. S. Melson and F. Schüth, J. Catal., 1997, 170, 46–53.
21. M. Bjørgen, S. Svelle, F. Joensen, J. Nerlov, S. Kolboe, F. Bonino, L. Palumbo, S. Bordiga and U. Olsbye, J. Catal., 2007, 249, 195–207.
22. G. Qi, Z. Xie, W. Yang, S. Zhong, H. Liu, C. Zhang and Q. Chen, Fuel Process. Technol., 2007, 88, 437–441.
23. T. Ino and S. Al-Khattaf, Appl. Catal., A, 1996, 142, 5–17.
24. N. Wang, W. Qian and F. Wei, J. Mater. Chem. A, 2016, 4, 10834–10841.
25. Y. Ma, D. Cai, Y. Li, N. Wang, U. Muhammad, A. Carlsson, D. Tang, W. Qian, Y. Wang, D. Su and F. Wei, RSC Adv., 2016, 6, 74797–74801.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra19035a

---