Silicoaluminophosphate-34 (SAPO-34) solid acid catalysts possessing a large chabazite (CHA) cage and a small eight-ring pore opening are widely used as important heterogeneous catalysts in various industrial processes because of their shape selectivity and controllable acidity [1-5]. However, the micropores always suffer from limited diffusion, resulting in fast coke deposition and a short catalytic lifetime [6]. To enhance SAPO-34 performance, methods for decreasing the crystal size and introducing secondary larger pores to enlarge the external surface and pore size of SAPO-34 structures have attracted growing attention from researchers [7-12].

Previous studies have proved that SAPO-34 catalysts with smaller particle size have advantages in mass transfer and catalytic lifetimes because of the higher proportion of accessible cages near the external surface [13-15]. However, with the smaller crystallite size and higher external surface area, these materials were very sensitive to hydrolysis by moisture under room conditions and suffered complete hydrolysis after months of storage [16]. Furthermore, nanosized catalysts are more difficult to recycle, which limits their application on a large scale. To further enhance mass transfer and reduce coke formation, the synthesis of thin zeolite layers is another strategy [13, 14]. Álvaro-Muñoz et al. [17] synthesized plate-like SAPO-34 nanocrystals by a microwave-assisted method. Nevertheless, these methods have disadvantages in their complex synthesis procedures. In addition to a reduction in particle sizes and the synthesis of thin zeolite layers, the introduction of mesopores or macropores into the bulk of SAPO-34 by preparing hierarchical SAPO-34 has been investigated extensively. In particular, the Ryoo research group developed a soft-templating synthesis method by introducing organosilane surfactants as mesopore directors in conventional zeolite synthesis compositions [18-22]. On the basis of this idea, by using [3-(trimethoxysilyl)propyl]-octadecyldimethylammonium chloride (TPOAC), an organosilane surfactant, as the mesopore director, hierarchical porous SAPO-34 can be obtained as an assembly of nanocrystallites intergrown into cubic micrometer-sized crystals [23-25]. Recently, SAPO-34 nanoaggregates were synthesized with the assistance of 3-piperazinepropyl- methyldimethoxysilane (PZPMS) [26]. Unfortunately, the addition of PZPMS and TPOAC is cost intensive. Therefore, it is essential to develop a low-cost strategy to synthesize SAPO-34 materials with large external surface areas and pore sizes. As far as we know, no strategies have been established to synthesize nanoaggregate SAPO-34 by using only the one-step hydrothermal method without adding other expensive pore-directing agents.

In this work, a nanoaggregate SAPO-34 catalyst with large external surface area and pore size was successfully synthesized without adding other expensive pore-directing agents. In addition, furfuryl alcohol (FAL), a biomass-derived chemical with a molecular size of about 0.53 nm, which is larger than the channels of traditional SAPO-34 (0.38 nm), was chosen as a model reactant to study the effect of external surface area and pore size on the catalytic alcoholysis performance of SAPO-34. Moreover, the alcoholysis of FAL is a promising strategy for utilizing renewable biomass resources [27] because ethyl levulinate (EL), the product from the alcoholysis of FAL, has potential applications in the transportation fuel sector and the chemical industry, for example, as a fuel additive or green solvent [28].

The materials used in the synthesis of SAPO-34 were pseudoboehmite (69 wt% Al₂O₃, Aluminum Corporation of China Limited, abbr. Chalco, Shandong Branch), phosphoric acid (85 wt%, Sinopharm Chemical Reagent Co. Ltd.), tetraethylorthosilicate (TEOS, 28 wt% SiO₂, Sinopharm Chemical Reagent Co. Ltd.), and tetraethylammonium hydroxide (TEAOH, 25 wt% aqueous solution, Shanghai Bai Ke Daily Chemical Co. Ltd.).

Furfuryl alcohol was purchased from Aladdin Chemical Reagent Co. Ltd. 2-(Ethoxymethyl)furan (EMF) was purchased from Bide Pharmatech Ltd. Benzyl alcohol, cyclohexane, n-butylamine, and ethanol were purchased from Sinopharm Chemical Reagent Co. Ltd.

All chemicals were used without further purification.

The typical synthesis procedure for nanoaggregate SAPO-34 was as follows. Pseudoboehmite (1.26 g) was first mixed with deionized water for 5 min. Subsequently, TEOS (1.11 g) and TEAOH (23.10 g) were added to the mixture, and violent stirring was continued for 1 h. Finally, H₃PO₄ (4.20 g) was added slowly. After further stirring of the mixture for 30 min, the obtained gel was transferred into a Teflon-lined stainless steel autoclave and heated at 180 ℃ for 24 h. The molar ratio of Al₂O₃: P₂O₅: SiO₂: TEAOH: H₂O in the SAPO-34 sol was 1.0: 2.0: 0.6: 4.0: 155.

For comparison, conventional cube-like SAPO-34 was also synthesized by using the method described elsewhere [29]. The molar ratio of Al₂O₃: P₂O₅: SiO₂: TEAOH: H₂O was 1.0: 2.0: 0.6: 4.0: 155, the same as that in the synthesized nanoaggregate SAPO-34. Unlike the synthesis of nanoaggregate SAPO-34, pseudoboehmite (1.26 g), TEOS (1.11 g), and H₃PO₄ (4.20 g) were first mixed and stirred for 2 h. TEAOH (23.10 g) was then added, and the mixture was stirred continuously for 4 h. The mixture was crystallized at 180 ℃ for 24 h in a Teflon-lined stainless steel autoclave under autogenous pressure.

The as-synthesized samples were obtained by filtration and washed thoroughly with distilled water. After being dried at 110 ℃ overnight, the solid products were calcined in air at 600 ℃ for 5 h to remove the templating agent.

Silicalite coating on SAPO-34 samples was prepared by a chemical liquid deposition method [30-32]. SAPO-34 coated with SiO₂ was produced with a molar ratio of 1.0 catalyst: 0.2 SiO₂: 50.0 cyclohexane. First, SAPO-34 (2.0 g) was suspended in cyclohexane (50 mL), then tetraethylorthosilicate (1.43 g) was added, and the mixture was stirred for 4 h. The mixture was dried at 120 ℃ and calcined at 540 ℃ for 3 h. The coating process was repeated twice. The obtained sample was denoted as SiO₂/SAPO-34.

Powder X-ray diffraction (XRD) patterns were recorded on a Bruker diffractometer by using Cu K_α radiation (40 kV, 40 mA). Pore and specific surface properties were acquired at -196 ℃ with a Micromeritics ASAP 2020 M sorption analyzer. All samples were outgassed at 300 ℃ for 24 h under vacuum to remove moisture and volatile impurities before the measurements. The surface area was calculated by the Brunauer-Emmett-Teller (BET) method. The pore volume and average pore diameter were obtained by the Barrett-Joyner-Halenda (BJH) method. Scanning electron microscopy (SEM) images were collected on a Nova NanoSEM 450 instrument by using an accelerating voltage of 3 kV. Chemical compositions were determined by inductively coupled plasma atomic emission spectrometry (ICP-AES) by using an Agilent 167nm-785nm/725 instrument. The ²⁷Al, ³¹P, and ²⁹Si magic-angle spinning (MAS) NMR spectra were obtained on a Bruker AVANCE III 500 spectrometer at a resonance frequency of 156.4 MHz. The chemical shifts were referenced to Al(NO₃)₃, tetramethylsilane, and 85% H₃PO₄ aqueous solution, respectively. The acidity of the catalysts was studied by NH₃ temperature-programmed desorption (NH₃-TPD) (PX200 instrument, Tianjin Golden Eagle Technology Limited Corporation). NH₃-TPD profiles were measured in the temperature range of 90-600 ℃ with a heating rate of 10 ℃ min^-1, and the desorbed ammonia was detected by using a thermal conductivity detector (TCD) at 110 ℃. The surface acid concentration was titrated with n-butylamine as the standard solution.

The alcoholysis reaction of furfuryl alcohol to ethyl levulinate was conducted in a 50 mL Teflon-lined stainless steel autoclave with a magnetic stirrer. In a representative test, FAL (0.15 g), catalyst (0.08 g), and ethanol (5.0 g) were added into the reactor. After being purged with nitrogen three times to remove the air inside at room temperature, the autoclave was charged with nitrogen to 0.5 MPa. It was then heated to 160 ℃ with magnetic stirring for a given time. At the end of the reaction, the reactor was immediately moved to a water bath to quench the reaction. The liquid was then separated from the solids by using a centrifuge, and the liquid was analyzed with an Agilent 7890 gas chromatograph equipped with an HP-5 capillary column connected to an flame ionization detector.

The XRD patterns of cube-like SAPO-34 and nanoaggregate SAPO-34 are shown in Fig. 1. The typical diffraction peaks of chabazite (CHA) topology at 2θ = 9.6°, 16.1°, 20.5°, 26.2°, and 30.9° appear in both patterns, confirming the formation of SAPO-34 zeolite [33]. Some differences in the peak width and relative intensities of these diffraction peaks can be observed between the patterns, suggesting that the samples have different particle sizes or morphologies [17]. This is also confirmed by the following SEM analysis.

As shown in the SEM images, cube-like SAPO-34 has crystals with a cubic shape and relatively uniform sizes of 1.5-2.0 μm (Fig. 2(A)). However, if the feeding order of H₃PO₄ and TEAOH is reversed, the morphology of the obtained SAPO-34 changes a lot: the SEM image shows spherical aggregates with a size of about 1 μm formed from many nano-sized crystals (Fig. 2(B)).

Calcined samples were analyzed by nitrogen adsorption-desorption to determine their textural properties. Both samples show steep uptake near P/P₀ = 0, which is characteristic for microporous materials (Fig. 3) [23]. Unlike the results for cube-like SAPO-34, a very small hysteresis loop exists at 0.4 < P/P₀ < 0.6 in the nanoaggregate SAPO-34 isotherms; this result is interpreted as capillary condensation in the inter-crystal pores [25, 34]. The data for pore volumes and surface areas calculated from the isotherms are collected in Table 1. It can be observed that nanoaggregate SAPO-34 exhibits a significantly higher external surface area (106 m²×g^-1) and mesopore volume (0.10 cm³×g^-1).

To determine the differences in the formation processes of nanoaggregate SAPO-34 and cube-like SAPO-34, several samples obtained after different crystallization times were characterized. Figures 4 and 5 show the XRD patterns and SEM images, respectively, of the solid products after different crystallization times. For nanoaggregate SAPO-34, after the first hour, the sample does not show any CHA characteristic diffraction peaks in the XRD pattern (Fig. 4) and exhibits spheroidal nanoparticle morphology with uniform size (Fig. 5(A)). After 3 h of crystallization, the sample has obvious peaks corresponding to the SAPO-34 structure, implying that crystallization has occurred. The particles are aggregated into spheres with a rough surface (Fig. 5(B)), which indicates that particle formation undergoes an assembly process. When the hydrothermal treatment time reaches 6 h, there is no significant change in the XRD result. However, we observed that the nuclei distributed on the external surface of the samples had grown upward and formed small islands (Fig. 5(C)) [35]. If the crystallization time is further increased to 12 h, the intensities of the XRD peaks increase, indicating the formation of a good crystalline structure, and the islands on the external surface have grown further and exhibit clear boundaries (Fig. 5(D)). However, the size of the sample still remains at about 1 μm, even after the longer crystallization time.

For comparison, Figs. 4 and 5 also show the XRD patterns and SEM images for cube-like SAPO-34 after different crystallization times. After 1 h and even after 3 h of crystallization, an amorphous phase without any regular shape is observed in the XRD results (Fig. 4) and SEM images (Fig. 5(E), 5(F)). With an increase in the crystallization time, the characteristic peaks of SAPO-34 are found and the plate shape morphology appears (Fig. 5(G)). With further prolongation of the crystallization time to 12 h, the intensities of the XRD peaks increase and the plate size becomes larger and thicker until cubes are finally formed (Fig. 5(H)).

Based on the above-mentioned results, a possible formation process for nanoaggregate SAPO-34 is illustrated in Scheme 1 (pathway I). First, after hydrolyzing in the alkaline solution, the aluminum and silicon precursors exist as Al(OH)₄^- and SiO₂(OH)^-, respectively; these species will probably combine with each other and form aluminosilicates with Si-O-Al bonds [36]. The aluminosilicates, with Si-O-Al bonds as the basic units of zeolite, will remain undissolved and will be inclined to produce more secondary units, including oligomers, amorphous particles, and even nanocrystallites, which can lead to a rapid and massive formation of nuclei [37-39]. This can be proved from the XRD and SEM results (Figs. 4 and 5). Next, self-assembly of the aluminosilicates by van der Waals attraction leads to the formation of spherical nanocrystals [40, 41]. Finally, the nanocrystals in the aggregate continue to grow upward to form nanoaggregate SAPO-34 with high crystallinity. For cube-like SAPO-34, the formation process is likely to follow the classical pathway mentioned in previous reports and shown in Scheme 1 (pathway II) [41, 42]. Unlike the precursors of nanoaggregate SAPO-34, the Al and Si sources exist as Al³⁺ and (OH)₃Si(OH₂)⁺, respectively, in acid solution [43]. After the addition of TEAOH, alumina in these colloidal suspensions will be deposited on the surface of the silica particles, which dramatically slows breakage of the Si-O-Si bonds. This can extend the time frame for reaching thermodynamic equilibrium or create a metastable state whereby the solution never reaches equilibrium prior to the onset of zeolite nucleation; that is, this can retard the rate of nucleation [39]. In this system, crystal growth occurs via a layer-by-layer mechanism that typically leads to well-defined crystals [41]. It can be seen that the different forms of precursors and rates of nucleation are the main reasons for the formation of the two types of SAPO-34 with distinct morphologies.

To probe the chemical environment of the framework atoms in both nanoaggregate SAPO-34 and cube-like SAPO-34, ²⁷Al, ³¹P, and ²⁹Si MAS NMR spectra were collected, as shown in Fig. S1. In a comparison of the two samples, nanoaggregate SAPO-34 shows more pentacoordinated and octahedral aluminum atoms, which are attributed to extra-framework aluminum species [25].

The acidity of the catalysts was evaluated by NH₃-TPD (Fig. 6). Both catalysts present two desorption peaks, one at low temperature (around 180 ℃) and the other at high temperature (around 380 ℃), which correspond to weak and strong acid sites, respectively [23]. The amounts of acid sites were determined from the peak area and are listed in Table 2. As seen from Table 2, the total acid amounts of nanoaggregate SAPO-34 (981 μmol×g^-1) and cube-like SAPO-34 (962 μmol×g^-1) are equivalent. With consideration that nanoaggregate SAPO-34 has a large external surface area, as determined from the N₂ adsorption-desorption result, the surface acid amount was also studied by n-butylamine titration (Table 2). Relative to the surface acid amount of cube-like SAPO-34 (22 μmol×g^-1), nanoaggregate SAPO-34 has greater surface acidity (263 μmol×g^-1), which is in accordance with the results for the external surface area.

Recognizing that cube-like SAPO-34 and nanoaggregate SAPO-34 have different external surface areas, pore sizes, and acid distribution, we decided to study the effect of these properties on the catalytic performance of the different SAPO-34 samples. Furfuryl alcohol (FAL), with a molecular size of about 0.53 nm, which is larger than the channels of traditional SAPO-34 (0.38 nm), was chosen as the model reactant. The alcoholysis reaction of FAL to form ethyl levulinate (EL) was studied with both cube-like SAPO-34 and nanoaggregate SAPO-34 samples, and the results are listed in Table 3. It can be seen that the conversion of FAL is very low (≤ 15.0%) with no catalyst or SiO₂. The SAPO-34 samples give full conversion of FAL, indicating that SAPO-34 can promote the alcoholysis of FAL. However, the yield of EL is obviously different with cube-like SAPO-34 and nanoaggregate SAPO-34: it reaches 19.9% over cube-like SAPO-34, whereas it is 74.1% over nanoaggregate SAPO-34. Unfortunately, the carbon balance in the FAL alcoholysis reaction is not very satisfactory. Humin, which cannot be detected by GC, may be produced because the solution is brown after the reaction. In addition, 2-(ethoxymethyl)furan (EMF) is the main side product observed with GC-MS. The existence of a great deal of EMF (48.1%) is the primary reason for the low EL yield over cube-like SAPO-34.

As we know, the alcoholysis of FAL to EL proceeds through the formation of the main intermediate EMF, which is then further converted into EL. The sizes of the related material molecules were acquired by using density functional theory with the generalized gradient approximation and the Perdew-Burke-Ernzerhof exchange-correlation functional, as implemented in the DMol3 modules of Material Studio 8.0. In Fig. S2, the value of the blue line denotes the distance between two atoms from different views, and molecule sizes can be obtained by adding the values for the radii of specific atoms [44]. It is easy to see that the sizes of EMF (0.55, 0.65, and 1.12 nm) and EL (0.31, 0.67, and 1.27 nm) are larger than that of the microporous channels in the SAPO-34 zeolite (0.38 nm). Therefore, it is reasonable to believe that the reaction mainly occurs on the external surfaces, including the mesoporous channels, of the SAPO-34 catalysts, rather than in the microporous channels. In addition, the external surface, including mesoporous channels with larger pore sizes, is helpful for the mass transfer of reactants and products.

To explore the role of EMF in the FAL alcoholysis reaction, the conversion of FAL over nanoaggregate SAPO-34 was conducted for different lengths of time, and the product distribution profile versus time is presented in Fig. 7. This shows that, during the first hour, the conversion of FAL is over 99% and the main products are EMF (43%) and EL (36%), with a small amount of DEP (7%). With a prolonged reaction time, the amount of EMF decreased gradually and that of EL correspondingly increased, indicating that EMF is the key intermediate in the FAL alcoholysis reaction, which is in agreement with previous reports [45, 46]. Notably, although complete conversion of FAL can be achieved within 1 h, the maximum EL yield is achieved after 6 h, with the decrease in the EMF yield. This result also implied that the conversion of EMF into EL is slow and rate-determining. According to the product distribution versus reaction time, the conversion of EMF is a key step in alcoholysis of FAL. Therefore, some experiments were performed with EMF as a substrate, and the results are shown in Table 4. There is still no significant difference between the EMF conversion with cube-like SAPO-34 and nanoaggregate SAPO-34. However, the yield of EL on nanoaggregate SAPO-34 (41.5%) was much better than that on cube-like SAPO-34 (12.7%). It can be seen that the large external surface is helpful for the conversion of FAL/EMF into EL.

In addition, to study the effect of the external surface area, the amount of cube-like SAPO-34 in the alcoholysis of FAL was increased to make the external surface area as large as that of the nanoaggregate SAPO-34. However, the catalytic performance with a triple dosage of cube-like SAPO-34, for which the EL yield was 37.7 %, was no better than that of nanoaggregate SAPO-34 (Table 3). This is likely because, although the amount of cube-like SAPO-34 has been expanded three times to makes the external surface area equivalent to that of nanoaggregate SAPO-34, the amount of surface acid of cube-like SAPO-34 (3 × 22 μmol×g^-1) is still much lower than that of nanoaggregate SAPO-34 (263 μmol×g^-1) (Table 2). In addition, nanoaggregate SAPO-34 has a larger pore size, which is beneficial for the diffusion of reactants and products.

To further study the effect of the external surface on the catalytic performance of SAPO-34 samples, the samples were coated with silica (SiO₂/SAPO-34) to reduce the external surface area. The weight ratio of silica to SAPO-34 is about 0.2:1. Figs. S3 and S4 show the XRD patterns and SEM images, respectively, of the SiO₂/SAPO-34 samples. After coating, there is no remarkable change to be found in the crystal structure or morphology. However, as indicated in Table 1, the external surface areas and mesopore volumes of both samples, especially SiO₂/nanoaggregate SAPO-34, are much lower than those of the uncoated samples. For SiO₂/nanoaggregate SAPO-34, in comparison with nanoaggregate SAPO-34, the external surface area and mesopore volume decrease from 106 to 45 m²×g^-1 and from 0.10 to 0.04 cm³×g^-1, respectively.

The catalytic performances of FAL alcoholysis with SiO₂/cube-like SAPO-34 and SiO₂/nanoaggregate SAPO-34 are shown in Table 3. It can be seen that the SiO₂/SAPO-34 samples exhibit much lower activity than those of the SAPO-34 samples, with the conversions of FAL dropping to 73.5 % with SiO₂/cube-like SAPO-34 and 51.4% with SiO₂/nanoaggregate SAPO-34, respectively. The yield of EL with cube-like SAPO-34 does not change significantly, going from 19.9% to 16.3%. However, with nanoaggregate SAPO-34, the yield of EL decreases substantially from 74.1% to 12.4% after SiO₂ coating.

Figure 6 and Table 2 show the total acid amounts and surface acid amounts of SiO₂/SAPO-34 samples determined by NH₃-TPD and n-butylamine titration, respectively. In a comparison with cube-like SAPO-34, the total acid amount of SiO₂/cube-like SAPO-34 decreases from 962 to 866 μmol×g^-1, which is 10.0% lower, and the surface acid amount falls from 22 to 13 μmol×g^-1. However, the surface acid amount for SiO₂/nanoaggregate SAPO-34 is significantly reduced compared to that for nanoaggregate SAPO-34, with a decrease from 263 to 28 μmol×g^-1. The decrease in the total acid amount over SiO₂/nanoaggregate SAPO-34 is owing to the change in the surface acid amount, which decreases from 981 to 759 μmol×g^-1 (22.6% lower). From the above results, we can conclude that the conversion of FAL and the yield of EL with SiO₂/SAPO-34 samples decrease as a result of the decline in the total acid amount and surface acid amount caused by the silica coating, which will cover the surface and block some pores.

Overall, nanoaggregate SAPO-34 has better catalytic performance for the FAL alcoholysis reaction than cube-like SAPO-34 largely because of its nanoaggregate morphology with a larger external surface area and mesopore volume, which leads to more surface acid and facilitates the accessibility of the reactant to the acid sites[17]. In addition, the effect of the amount of external surface acid is more important than that of the pore size for the catalytic alcoholysis reaction of FAL over SAPO-34 samples.

The stability of the SAPO-34 catalysts for the alcoholysis of FAL to form EL was tested. To better exhibit the real stability of the catalyst, the reaction time was fixed at 1 h and the dosage of SAPO-34 was 0.04 g, both of which values are lower than those in Table 3. As shown in Fig. 8, the deactivation of cube-like SAPO-34 is obviously observed after two runs, with FAL conversion reduced by 42%. However, nanoaggregate SAPO-34 maintains similar FAL conversion after two runs. Moreover, the FAL conversion only decreases by 23% after five cycles (Fig. S5), demonstrating that the stability of SAPO-34 has been enhanced dramatically by improving the catalyst's external surface area. In a comparison with those of fresh nanoaggregate SAPO-34, the XRD patterns and SEM images of the used nanoaggregate SAPO-34 show no obvious change in the crystal structure and morphology (Figs. S6 and S7), which proves the structural stability of nanoaggregate SAPO-34. Moreover, used nanoaggregate SAPO-34 was characterized by thermogravimetric (TG) analysis after five cycles (Fig. S8). The weight loss curve indicates that coke is deposited on the zeolite. After regeneration of the used nanoaggregate SAPO-34 catalyst, which was conducted in air at 600 ℃ for 3 h, the activity basically recovered. According to the XRD and TG results, the deactivation of nanoaggregate SAPO-34 is caused by coke deposition. Moreover, used nanoaggregate SAPO-34 and used cube-like SAPO-34 were characterized by TG analysis after one reaction (Fig. S9). From the weight loss curve, the coke amount on the used nanoaggregate SAPO-34 (2.8 wt%) is lower than that on the used cube-like SAPO-34 (4.3 wt%). This may be because the nanoaggregate SAPO-34 sample has a larger external surface area than that of cube-like SAPO-34, as a result of the aggregation of the nanocrystals, which is helpful for decreasing coke deposition.

Nanoaggregate SAPO-34 was synthesized by a low-cost hydrothermal method without adding other expensive pore-directing agents. Relative to cube-like SAPO-34, nanoaggregate SAPO-34 has a larger external surface area and mesopore volume, which leads to more external surface acid sites and facilitates the accessibility of the reactant to the acid sites. When the SAPO-34 catalysts were applied to the catalytic alcoholysis of FAL to form EL, nanoaggregate SAPO-34 exhibited good catalytic activity; the EL yield with nanoaggregate SAPO-34 was almost fourfold higher than that with cube-like SAPO-34. The importance of the external surface is further confirmed by a comparison of the catalytic performances of nanoaggregate SAPO-34 and SAPO-34 coated with silica. The results showed that the conversion of FAL and the yield of EL with SiO₂/SAPO-34 decreased quickly because the silica coating covers the surface and blocks some of the pores, which reduces the amount of surface acid sites.