Two-Step Dry Gel Method Produces MgAPO-11 with Low Aspect Ratio and Improved Catalytic Performance in the Conversion of Methanol to Hydrocarbons

: In this article, the synthesis, characterization and catalytic performance of three MgAPO-11 catalysts with distinct crystal morphologies (sunﬂower, ball and candy) are presented. Among the three samples, the candy-like MgAPO-11-C, with high crystallinity and uniform particle size (of about 1 µ m), was synthesized for the ﬁrst time by using a unique two-step dry gel method. Despite the similar acid strength of the three samples, the different and distinct morphologies of the catalysts resulted in very different methanol-to-hydrocarbons (MTH) performances. In particular, the candy-like MgAPO-11-C presented the best MTH performance with the highest total conversion capacity (4.4 g MeOH · g catalyst − 1 h − 1 ) and the best selectivity to C 5+ aliphatics (64%).


Introduction
Despite decreasing crude oil reservoirs and pressing environmental concerns, the global demand for energy and petrochemical products is still increasing [1][2][3]. This situation calls for alternative routes for the manufacture of such products [4][5][6]. The conversion of renewable and recycled C/H sources into high-value products is considered a promising path towards meeting this increasing demand. The methanol-to-hydrocarbons (MTH) reaction is the final step in such a process [7]. Zeolites and zeotypes, known as shape selective materials, are key for the MTH reaction, due to their ability to distinguish between reactants, products and reaction intermediates, based on their molecular size. Mechanistic studies indicate that the MTH reaction proceeds by a "hydrocarbon pool mechanism" as proposed by Dahl and Kolboe and built upon by Svelle et al. in the "dual cycle" mechanism [8,9]. The "Hydrocarbon pool mechanism" can be divided into two parts: one is the "alkene cycle" and the other is the "aromatic cycle". Ethene is produced primarily from the "aromatic cycle", while propene and higher alkenes may be formed from both cycles. This is due to the promotion of distinct reaction intermediates and reactions by each topology [7,10,11]. Furthermore, diffusion restrictions contribute substantially to product sieving into the bulk phase. Hence, the product distribution from a given feedstock may be tuned by using catalysts with different topology, such as the conversion of methanol to gasoline over ZSM-5 and methanol to olefins over SAPO-34 [12,13].
Recently, focus has been set on developing new molecular sieves for the MTH reaction, such as the one-dimensional zeolites ZSM-22 and ZSM-23 [14][15][16][17]. ZSM-22 and ZSM-23 contain one-dimensional non-intersecting 10-ring channels with diameters of 0.46 nm × 0.57 nm [15,18]. As a result of this unique channel structure, the "aromatic cycle" of the MTH reaction is highly restricted, and both ZSM-22 and ZSM-23 show a high selectivity for C 5 -C 6+ alkenes, free of aromatics [17]. Hence, ZSM-22 and ZSM-23 based MTH products might help meet current and future petrochemical demands and could be suitable for gasoline production via hydrogenation. The 1D 10-ring zeolites are prone to rapid deactivation by the formation of aromatic compounds that block the access to pores containing the active sites [18]. Recent research demonstrated that a smaller particle size, characterized by a low aspect ratio, dramatically enhances the catalytic lifetime of ZSM-23 [15].
In another line of work, studies of the 1D 12-ring AFI structure showed that the addition of the isostructural AlPO-5 to a SSZ-24 zeolite increased its MTH turn-over number by facilitating methanol dehydration to dimethyl ether (DME) [19]. Furthermore, studies of MAPO-5 (M = Mg, Co, Zn, Si, Ti, Zr) suggested that the more strongly acidic MgAPO-5 analogue favors the alkene cycle over the arene cycle compared to SAPO-5, leading to enhanced propene selectivity over 21]. Encouraged by these results, we decided to explore a 1D 10-ring MgAPO material, MgAPO-11, as a catalyst for the MTH reaction.
The zeotype AlPO 4 -11 material has an AEL topology and consists of alternating, (AlO 4 ) − and (PO 4 ) + tetrahedra, resulting in a neutral framework [22][23][24]. In MgAPO-11, Al is partially substituted by Mg in the AlPO 4 -11 lattice, thereby introducing Brønsted (B) acid sites [25,26]. The diameters of the major and minor axes of the 10-ring channels in MgAPO-11 are 0.65 nm and 0.43 nm. MgAPO-11 has attracted a great deal of attention due to its excellent performance in the hydroisomerization of n-alkanes, which produces low freezing point diesel, high octane number gasoline, and high viscosity lubricants [27][28][29]. However, to the best of our knowledge, the potential of MgAPO-11 as a catalyst for the MTH reaction is yet to be explored. Furthermore, most of the papers on MgAPO-11 have focused on MgAPO-11 with a large particle size along the single channel structure, which is less suitable for the MTH reaction [25,26,30]. Recently, Shu et al. successfully synthesized hierarchical MgAPO-11 molecular sieves by using the epoxide-mediated solgel method, however, the synthesis uses propylene oxide and harsh conditions (ice-cooled conditions) [27]. Searching for a simpler synthesis procedure to obtain small, hierarchical MgAPO-11 crystals, we found inspiration in two contributions from the Chen and Rimer groups, respectively, who used ethanol as a growth modifier to obtain zeolite L with a low aspect ratio [31,32].
In this work, a unique two-step dry gel method, using ethanol as a growth modifier, has been developed to synthesize MgAPO-11 with a low aspect ratio and hierarchical structure. The physical characteristics and catalytic properties of this material for the MTH process was compared to materials obtained by alternative synthesis routes. The effect of crystal morphology on the catalyst lifetime was profound. The best catalyst synthesized by the unique two-step dry gel method showed a conversion capacity seven-fold higher than the other samples that were synthesized.

Characterization
Scanning electron microscope (SEM) images (Figures 1 and S1) of the three MgAPO-11 samples clearly show three distinct morphologies (and particle sizes) produced by the three different synthesis methods used. These MgAPO-11 samples exhibited crystal morphologies reminiscent of sunflowers (MgAPO-11-S), balls (MgAPO-11-B), and candy (MgAPO-11-C). Close inspection of the SEM images reveals that MgAPO-11-S and MgAPO-11-B particles consist of bundles of needle-like crystals with a high aspect ratio. The MgAPO-11-S sunflowers are composed of approximately 5 µm long needles, while the MgAPO-11-B balls consist of approximately 2 µm long needle-like particles and some amorphous particles. In contrast, the 1 µm particles of MgAPO-11-C consisted of singular crystals with uneven surfaces and an aspect ratio close to unity. The elemental ratios of Mg and Al (Mg/Al) were calculated from scanning electron microscope-X-ray energy dispersion spectra (SEM-EDS) ( Table 1). The Mg/Al ratio of the three samples varied from 0.06 to 0.08, corresponding to the 305-399 µmol Mg/g sample.   Powder X-ray Diffraction (PXRD) patterns of the three MgAPO-11 samples (Figure 2a) were consistent with the AEL structure. MgAPO-11-B showed a lower peak intensity compared to the other two samples, probably due to the existence of an amorphous phase [33]. MgAPO-11-C displayed the highest peak intensity among the three samples, indicating that it had the highest crystallinity [34].
The local atomic environment of P in the three MgAPO-11 samples was investigated using solid-state NMR analysis. The 31 P magic angle spinning nuclear magnetic resonance ( 31 P MAS NMR) spectrum ( Figure 2b) of all three samples exhibited an intense peak at −30 ppm and a shoulder peak at −24 ppm, and we have attributed this to the tetrahedral P(4Al) environment and P(3Al, 1Mg), respectively [27,35]. In addition, there is a broad peak at 0 to −20 ppm for MgAPO-11-B, and this broad peak may be assigned for P(2Al, 2Mg) and P(1Al, 3Mg) [36]. These results further confirmed the successful introduction of Mg 2+ into the framework.  (Table 1). These results showed that, among all three MgAPO-11 samples, MgAPO-11C had the largest number of weak and strong B and L acid sites accessible for pyridine. The low number of B sites available to pyridine compared to the total amount of Mg in the samples could be linked to extra-framework Mg, as well as to restricted diffusion of pyridine into the crystals. Since the size of pyridine is close to the size of the AEL pores, the latter effect is probably important. The difference in acid strength distribution between ammonia adsorption and pyridine adsorption ( Figure 3, Table 1) points in this direction. The N 2 adsorption-desorption isotherms ( Figure S2A) for MgAPO-11-C exhibited behavior typical for a type-IV isotherm with a H4 type hysteresis loop. This is indicative of a material with uniform slit-like mesopores. In comparison, MgAPO-11-B and MgAPO-11-S exhibited a type-IV isotherm with a H3 type hysteresis loop behavior, indicating an uneven distribution of slit-like mesopores, as confirmed by pore size distribution plots ( Figure S2B) [37]. Hence, micropores and mesopores co-exist in all three samples [38].
The textural properties of the three MgAPO-11 samples are listed in Table 2 and highlights the advantageous properties of MgAPO-11-C that were achieved using this twostep dry gel method. MgAPO-11-C not only showed the largest Brunauer-Emmett-Teller (BET) surface area (254 m 2 ·g −1 ) compared with MgAPO-11-S (95 m 2 ·g −1 ) and MgAPO-11-B (187 m 2 ·g −1 ). It was also shown to be larger than that achieved previously for MgAPO-11 (175 m 2 ·g −1 ) synthesized by a traditional dry-gel method [27]. In addition, MgAPO-11-C showed the highest external specific surface area (177 m 2 ·g −1 ) and highest hierarchical factor (0.093) of all three samples. This indicates that this two-step dry gel method is able to produce small and highly mesoporous MgAPO-11 particles not seen with the traditional dry-gel methods and the other two methods reported in this work. The micropore volume of the MgAPO-11-C (0.037 m 3 ·g −1 ) was much larger than that of MgAPO-11-S (0.018 m 3 ·g −1 ) and MgAPO-11-B (0.018 m 3 ·g −1 ), yet smaller than MgAPO-11 synthesized by traditional dry-gel method (0.06 m 3 ·g −1 ) [27]. This result indicates that the dry-gel method is beneficial for the synthesis of a mesoporous MgAPO-11, without the loss of the microporous structure. a Brunauer-Emmett-Teller (BET) surface area; b Total pore volume at P/P 0 = 0.990; c Micropore volume from t-plot; d V meso = V Total − V mic ; e Hierarchy factor = (V mic /V Total ) × (S ext /S BET ); f The content of coke was analyzed by Thermo-gravimetric (TG) up to 800 • C with a ramp rate of 10 • C/min; g R coke = coke amount/reaction time.

Catalytic Activity
The catalytic performance tests of the three MgAPO-11 catalysts were carried out at atmospheric pressure and 450 • C, with a weight hourly space velocity (WHSV) of 0.5·h −1 (Figure 4). The three catalysts displayed a wide variation in lifetime (Figure 4a). This is quantified further by comparing the total conversion capacities ( Table 2), defined as the total amount of methanol converted into hydrocarbons until complete deactivation [17]. MgAPO-11-C showed the highest total conversion capacity of all three samples with 4.4 g MeOH ·g cat −1 , six-to seven-fold larger than MgAPO-11-S and MgAPO-11-B, respectively (Table 2). In addition to morphology and crystal size, the total conversion capacity also varied with the Mg content of the catalysts. For MgAPO-11-C samples synthesized with an increasing Mg content ( Figures S3-S8), the total conversion capacity also increased from 0.2 to 4.4 g MeOH ·g cat −1 . With the exception of MgAPO-11-S, a linear relationship was observed between the amount of Brønsted acid sites accessible to pyridine and the total conversion capacity of each sample ( Figure S9). All tested MgAPO-11 samples displayed a similar slope in their deactivation curves, which was typically seen when the methanol conversion had dropped to 80%. At this stage, the activity drop was close to 20 conversion points per hour. Overall, the test results and characterization data suggest that the main reason for the difference in conversion capacity is the number of accessible B acid sites (Tables 1 and S1, Figure S9). A similar conclusion has previously been reached for zeolites with ZSM-22 and ZSM-23 topologies [15,16,39]. In order to further explore the deactivation process of the three catalysts, the amount of the deposited coke of the three deactivated samples was analyzed after the MTH reaction. The weight losses for MgAPO-11-S, MgAPO-11-B and MgAPO-11-C were 3.74%, 4.79% and 7.22%, respectively ( Table 2 and Figures S10 and S11), which are obtained by burning the retained coke species. Considering the different reaction times for the three catalysts, the average coke formation rate was calculated by using the equation: R coke = coke amount/reaction time. The R coke value of MgAPO-11-C (0.584) is much lower than that of MgAPO-11-S (0.733) and MgAPO-11-B (1.362), which agrees with the better performance and longer lifetime of MgAPO-11-C. Considering next the product selectivity of the MgAPO-11 samples, C 5+ aliphatics was the main product obtained over all samples (Figure 4b-d). Focusing first on MgAPO-11-C, it displayed a selectivity pattern where: C 5+ >> C 3 > C 4 >> C 2 > C 1 . The maximum C 5+ selectivity obtained was 64%. With progressive deactivation, the aromatics selectivity rose to become the second most abundant product, followed by methane (C 1 ). Considering the small pore size of MgAPO-11, it is likely that aromatic products were formed either at the micropore pore mouth, in the mesopores or at the external surface of MgAPO-11-C. Focusing next on MgAPO-11-S, its selectivity pattern was very similar to that of MgAPO-11-C, although with somewhat higher selectivity to C 2 -C 4 , giving a maximum C 5+ aliphatics selectivity of 55 %. Considering, finally, MgAPO-11-B, C 5+ aliphatics was still the main product (maximum 33% selectivity), yet with even higher selectivity towards C 2 -C 4 , and to methane and aromatics with progressive deactivation, compared to the other two samples. To elaborate on the selectivity variations, it is necessary to turn to the MTH reaction scheme, as described in a recent review [1]. Methane, ethene and propene are primary products of the initiation reactions leading to the first C-C bond formation. During steady-state hydrocarbon pool operation, propene and butenes are formed from the cracking of higher hydrocarbons, and from ethene/propene methylation. Ethene is mainly formed from dealkylation of polymethylated benzene (and naphthalene) intermediates. Considering the prior work on 1D 10-ring zeolites (ZSM-22, ZSM-23), aromatic products were found to be virtually inert to further reaction, and even the cracking of higher hydrocarbons is severely restricted by the small pore size [40,41]. With the similar pore size of MgAPO-11 and those zeolites, it is likely that cracking and dealkylation reactions are also restricted in MgAPO-11. Although we have no direct evidence, it is tempting to suggest that the higher selectivity to C 1 -C 4 products in the MgAPO-11-B and MgAPO-11-S materials may be due to a higher relative abundance of initiation reactions compared to hydrocarbon pool propagation reactions, in these materials compared to MgAPO-11-C. This suggestion would be in line with the lower conversion capacity observed for MgAPO-11-B and MgAPO-11-S.
The aim of this study was to explore the potential of MgAPO-11 as a catalyst for the MTH reaction. As already pointed out, this material has similar pore size and dimensionality to ZSM-22 and ZSM-23. Considering the product selectivity of MgAPO-11, it is also very similar to ZSM-22 and ZSM-23, with the aliphatics selectivity decreasing in the order: C 5+ >> C 3 > C 4 >> C 2 > C 1 . A direct comparison between the materials is complicated by the fact that ZSM-22 and ZSM-23 selectivity results are reported at 400 • C, while MgAPO-11 was tested at 450 • C. However, keeping in mind that the comparison is rather qualitative, the maximum C 5+ aliphatics selectivity reported for ZSM-22 is around 70%, somewhat higher than the 64% maximum selectivity reported for MgAPO-11-C in this work. Furthermore, the maximum conversion capacity of ZSM-22 reported at 450 • C was 12 g MeOH ·g cat −1 [39], 2.7 times that reported for MgAPO-11-C here. Considering the results reported in this work and the deactivation pattern of 1D 10-ring structures during the MTH reaction, the difference is probably strongly related to the contact time and number of accessible Brønsted acid sites in the two materials [18,39].

Synthesis of MgAPO-11-S
For the synthesis of MgAPO-11-S, 3.0 g of boehmite was dissolved in 16.0 g of deionized water and stirred for 1 h at 25 • C. Then 5.1 g of H 3 PO 4 was added, followed by the addition of 0.6 g of Mg(NO 3 ) 2 ·6H 2 O and stirring at 25 • C. After 4 h, 3.4 g of di-npropylamine (DPA) was added and the gel stirred for another 24 h at 25 • C. Finally, the gel was transferred into a 45 mL PTFE-lined stainless-steel autoclave and pre-crystalized at 100 • C for 24 h with rotation (25 rpm), followed by crystallization at 200 • C for an additional 24 h with rotation (25 rpm). The resultant solid was recovered by filtration, washed three times with deionized water and dried at 100 • C. The template was subsequently removed by calcination at 600 • C for 4 h in air.

Synthesis of MgAPO-11-B
MgAPO-11-B was synthesized based on the two-stage hydrothermal crystallization method previously reported by Zhang et al. with Mg replacing Si in the synthesis [42]. In detail, the first stage involves the addition of 4. The unique point of the current synthesis method compared to those used in the prior literature is that a two-step dry gel method with additional controlled solvent evaporation was used to achieve the small and hierarchical MgAPO-11 particles. Control of the solvent evaporation process is important; as ethanol needs to be evaporated, although only to an extent where some ethanol is left for the second step crystallization to work. When ethanol was replaced with water, larger crystals with a high aspect ratio were obtained.

Catalyst Characterization
Siemens Bruker D500 in Graff Brentano geometry with Cu Kα radiation (λ = 1.5406 Å) was used to conduct the PXRD measurements. The size and morphology of samples were evaluated by using a Hitachi SU 8230 field-emission environmental scanning electron microscope (FESEM). The textural properties of the calcined samples were tested by N 2 adsorption-desorption measurements at −196 • C (Belsorp-mini II). The specific surface areas of the samples were analyzed by the BET method and the surface areas and pore volumes of the samples were calculated by the t-plot method. The strength and distribution of acid sites were analyzed by NH 3 -TPD. Py-IR adsorption spectra were obtained on a Nicolet 5700 spectrometer with a resolution of 1 cm −1 . 31 P MAS NMR spectroscopy characterization was carried out on a Bruker DSX 500 MHz NMR spectrometer. TG analyses were performed on a Mettler Toledo instrument, in which each sample was heated between 40 and 800 • C at a ramp rate of 10 • C/min in an air flow.

Catalytic Testing
The MTO performance testing of the samples was performed in a continuous downflow bed U-shaped quartz reactor (4 mm in internal diameter). The test rig was equipped with Brooks mass flow controllers for gas feeding at atmospheric pressure. Then, 0.2 g of a calcined MgAPO-11 catalyst was loaded into the reactor and pre-treated in a pure O 2 flow of 25 mL/min at 550 • C for 2 h before each run. The reactor was subsequently cooled down to 450 • C in an He flow. Methanol was fed to the reactor by bubbling an He flow (3.3 mL/min) through a methanol saturator. The saturator temperature was kept at 35 • C, giving rise to a methanol partial pressure of 280 mbar, corresponding to a methanol flow of 1.3 mL/min and a WHSV of 0.5 h −1 . The reaction products were analyzed using an online gas chromatograph (Agilent 6890 A) with a flame ionization detector (FID) and a capillary Rtx-DHA-150 column (60 m × 0.53 mm × 3 µm).

Conclusions
Three MgAPO-11 samples with different morphologies were successfully synthesized, including a candy-like MgAPO-11-C with a low aspect ratio obtained by using a novel twostep dry gel method. The three MgAPO-11 samples had similar acid strength distribution, however, they had dramatically different crystal morphologies and number of accessible Brønsted acid sites. The catalytic performance of the three MgAPO-11 materials for the MTH reaction was evaluated at 450 • C. The candy-like MgAPO-11-C with its smaller particle size and higher number of accessible Brønsted acid sites showed the highest stability and selectivity to C 5+ aliphatics.
In summary, it is demonstrated that the size and shape of MgAPO-11 particles can heavily influence their stability and selectivity to C 5+ aliphatics in the MTH process. This article also presents a novel route to prepare MgAPO-11 catalysts.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/catal12040413/s1, Table S1: Amount and distribution of acid sites in MgAPO-11 with different Mg contents; Figure S1: SEM images of the three MgAPO-11 samples with distinct morphologies; Figure S2: Nitrogen adsorption-desorption isotherms of the three MgAPO-11 samples with distinct morphologies; Figure S3: SEM images of MgAPO-11-C samples with different Mg contents; Figure S4: XRD patterns of the MgAPO-11-C samples with different Mg contents; Figure S5: NH 3 -TPD profiles of the MgAPO-11-C samples with different Mg contents; Figure S6: Py-IR spectra of MgAPO-11 with different Mg contents; Figure S7: N 2 adsorption and desorption isotherms of the MgAPO-11-C samples with different Mg contents; Figure S8: Methanol conversion as a function of time on stream of the MgAPO-11-C samples with different Mg contents; Figure S9: Methanol conversion capacity versus accessible Brønsted acid sites; Figure S10: TG curves of the three MgAPO-11 samples with distinct morphologies after reaction; Figure S11: Photos of the three MgAPO-11 samples after reaction.

Conflicts of Interest:
The authors declare no conflict of interest.