Acrylic acid (AA) and methyl acrylate (MA), the two extremely important industrial chemicals, are widely applied in the production of paintings, coatings, carbon fibers, and adhesives. They are mostly produced by a two-step oxidation reaction of propylene [1-3]. This route of oxidation by air/oxygen incurs a risk of over-oxidation of propylene and the desired products, which limits their widespread application on a large scale. Therefore, a novel and green route for MA and AA synthesis is highly demanded. Recently, the route to produce MA and AA through a one-step aldol condensation reaction of formaldehyde (FA) and methyl acetate (MAc) has been attracting substantial interest from both academic and industrial communities because of its simplicity and common feedstocks that can be derived readily from natural gas, coal, and biomass [4, 5].

Aldol condensation reaction, which is catalyzed by acid/base catalysts, can readily occur over either an acid (or a base) or an acid-base bifunctional catalyst. Previous studies have mainly focused on aldol condensation catalyzed by a variety of cesium supported with SiO₂ or SBA-15 [6-9], by acid–base bifunctional catalysts such as V₂O₅-P₂O₅ binary or V₂O₅-P₂O₅-SiO₂ ternary systems [10-13], or by alkali metal oxide-supported acidic catalysts [14-16]. However, there are limited experimental data available for evaluating the performance of solid acid catalysts in the aldol condensation reaction in the gaseous phase. Zeolite catalysts, as a type of solid catalysts being comprehensively utilized for more than 40% industrial processes in petroleum and petrochemical fields [17], show significant potential in aldol condensation. In our previous study [18], several types of zeolites, including HY, H , HMOR, HZSM-5, and HZSM-35, were first used as aldol condensation catalysts for synthesizing MA and AA with high activity, which suggested that zeolite is a catalyst candidate for the aldol condensation reaction of FA and MAc.

It is well known that two types of acids, Brӧnsted and Lewis acids, exist in zeolites, especially in silica-alumina zeolites. However, the issue of whether the aldol condensation reaction occurs over Lewis acid sites, Brӧnsted acid sites, or both has not been resolved yet. Jeong et al. [19] reported the results from in situ Fourier-transform infrared spectroscopy (FT-IR) studies of aldol condensation reactions of aldehydes with AlPO₄ zeolite as a catalyst. They found that Lewis acid was the active site for the aldol condensation reaction due to the lack of a bridging hydroxyl group in the AlPO₄ zeolite catalyst. Panov et al. [20] conducted aldol condensation reactions over nanosized amorphous alumina catalysts containing different amounts of Lewis acid sites and found that the rate of aldol condensation reaction of acetone decreased with the loss of Lewis acid sites. Dumitriu et al. [21] synthesized a series of MFI zeolites by isomorphous substitution of Me³⁺ for silicon and evaluated them in aldol condensation reactions of acetaldehyde and formaldehyde. They found that both Brӧnsted and Lewis acids were involved in the reaction and that a higher Lewis acid concentration favored a better selectivity for MFI zeolite catalysts. However, Kikhtyanin et al. [22] proved that solid catalysts with exclusive Lewis acid sites possessed substantially lower activity in the aldol condensation of furfural and acetone with MOFs as acidic catalysts, and suggested that Brӧnsted acid contributed more to the aldol condensation reaction than Lewis acid. Kikhtyanin et al. [23] also reported that two HBEA samples with the same Brӧnsted acid concentration but different Lewis acid concentrations exhibited approximately the same activity. They believed that the aldol condensation reaction proceeded with the participation of Brӧnsted acid rather than Lewis acid.

In our previous studies [16, 18], HZSM-35 zeolite was found to perform very well in the aldol condensation reaction of MAc and formaldehyde, and its acid site was found to be indispensable for triggering the reaction. To the best of our knowledge, no information is available yet on the effect of acidic properties of zeolites on the aldol condensation of formaldehyde and MAc to prepare MA and AA. Therefore, here, we study in detail the acidic property of HZSM-35 and prepare ZSM-35 with different concentrations of Brӧnsted acid through the sodium ion-exchange process without altering the concentration of Lewis acid. We then evaluate the catalytic performance of ZSM-35 zeolites for the aldol condensation reaction of formaldehyde and MAc.

The regents used included dimethoxymethane (DMM) (AR, 98%, Alfa Aesar (China) Chemicals Co., Ltd.) and MAc (AR, 98%, Sinopharm Chemical Reagent Co., Ltd., China). Na-ZSM-35 (SiO₂/Al₂O₃ = 79) was purchased from Shanghai Novel Chemical Technology Co., Ltd., China. Zeolite was calcinated at 823 K in air for at least 5 h to remove the retained organic template, and then converted into an H-form zeolite before use by conducting ion exchange thrice at 353 K for 6 h in 1 mol/L NH₄NO₃ solution, followed by washing with deionized water, drying at 373 K overnight, and calcination at 823 K for 4 h. To obtain zeolites with different sodium ion weight percentages, the H-form zeolite powder began to be exchanged with NaNO₃ solution (0.1 mol/L) at 323 K with a solid-to-liquid ratio of 1/20 for different times. The filter cake was dried at 373 K for 12 h. At last, we obtained zeolite with different sodium ion-exchange degrees. The sample was indicated as Na-Z-x, where x indicates the molar ratio of sodium element to aluminum element in the zeolite characterized by X-ray fluorescence (XRF) results.

FT-IR spectroscopy was conducted at a spectral resolution of 4 cm^–1 on a Bruker Tensor 27 FT-IR spectrophotometer equipped with a mercury-cadmium-telluride detector, which was sensitive to –OH group vibration. The sample was pressed into a self-supporting disk with a diameter of 13 mm. Then, the disk was put into a quartz cell, which was connected to a vacuum system and sealed with CaF₂ windows and then heated to 723 K for at least 4 h to remove the retained water before collecting the spectra. The pyridine or acetonitrile adsorption was performed by exposing the preheated disk to its vapor. The semi-quantitative analysis of the concentration of Brӧnsted or Lewis acid is shown as below [24]:

The solid-state NMR experiments were carried out on a Bruker AvanceIII 600 spectrometer equipped with a 14.1 T wide-bore magnet. The bulk acidity of zeolite was determined by temperature-programmed desorption of ammonia (NH₃-TPD) on a Micromeritics AutoChem 2920 instrument. The sample (0.20 g) was loaded in a U-shaped microreactor and preheated at 823 K for 0.5 h under helium atmosphere. After cooling to 373 K, the sample was saturated with ammonia, followed by purging with helium to remove the physically adsorbed ammonia molecule. Ammonia desorption was conducted in helium flow (30 mL/min) by heating from 373 to 923 K at a rate of 10 K/min and measured by a thermal conductivity detector.

The crystallinity of the samples was characterized by a PANalytical X'Pert PRO X-ray diffraction (XRD) meter with Cu-K_α radiation (λ = 1.51059 Å) at 40 kV and 40 mA. The chemical composition of zeolite was determined by Philips Magix-601 XRF. Nitrogen adsorption-desorption isotherms were obtained on a Micrometrics ASAP 2020 system at 77 K.

Aldol condensation of formaldehyde with MAc was carried out at 623 K with 0.5 g zeolite (20–40 mesh) in a fixed bed. Here, DMM was applied as the source of formaldehyde. The system pressure was set at 3 MPa. DMM and MAc precursors in two separated stainless-steel tubes held at 293 K were bubbled into the reaction tube by 30 mL/min N₂ flow, respectively. The products were analyzed using an online gas chromatograph equipped with a flame ionization detector connected to an FFAP capillary column. The yield of MA and AA was defined as moles_MA+AA/moles_{MAc fed} and S_MA+AA was calculated using the moles_MA+AA/moles_MAc consumed.

FT-IR spectroscopy is applied to characterize the acid property of HZSM-35 zeolite, as shown in Fig. 1. The spectrum of HZSM-35 zeolite in Fig. 1(a) consists of two distinct bands at 3745 cm^–1 due to the terminal silanol groups and at 3599 cm^–1 due to the bridging hydroxyl groups (Brӧnsted acid sites) [25-27]. The relatively high intensity of silanol groups at 3745 cm^–1 shows that HZSM-35 zeolite is high-siliceous, which agrees with the XRF result indicating a high SiO₂/Al₂O₃ ratio of 79. In addition, a broad band around 3650 cm^–1 is assigned to OH groups adjacent to extra-framework aluminum species. A close examination reveals that hydroxyl IR bands are asymmetric, which is particularly evident from the first derivative of the spectrum. The asymmetry at 3745 cm^–1 can be attributed to the presence of terminal silanol groups and to their hydrogen bonding interaction, while that at 3599 cm^–1 is associated with the Brӧnsted acidic bridging hydroxyl moieties vibrating in channels or cages of different sizes [28].

The ²⁹Si MAS NMR spectrum of HZSM-35 zeolite in Fig. 2(a) shows three signals at –105.7, –112.1, and –116.6 ppm, which are assigned to Si(1Al) T_A, the superposition of Si(1Al) T_A and Si(0Al) T_B, and Si(0Al) T_B, respectively [29-31]. The high ratio of Si(0Al) and Si(1Al) contributions indicates the prevalence of the Si–O–Si bond in the absence of neighboring Al. Fig. 2(b) shows the ²⁷Al MAS NMR spectrum of HZSM-35 zeolite used here, where the signal at 54.5 ppm is attributed to tetrahedrally coordinated framework aluminum atoms. The signal at 0 ppm indicates the presence of a few octahedrally coordinated extra-framework aluminum species, acting as Lewis acid sites, which is consistent with the FT-IR observation in Fig. 1.

Fig. 3 shows the deconstruction of the bridging OH band in the IR spectra of HZSM-35 according to Zholobenko et al. [28]. A good fit is obtained by decomposition based on three peaks. The 3610 cm^–1 band can be attributed to Si–O(H)–Al groups in 10-member rings, amounting to 23% of Si–O(H)–Al groups. The intense band at 3600 cm^–1 can be assigned to the bridging hydroxyl groups in cages at the intersection of 8- and 6-ring channels, amounting to 51% of the bridging hydroxyl groups. The band at 3588 cm^–1 can be attributed to Si–O(H)–Al groups in 8-member rings. Such an assignment coincides with the republished work. For instance, the initially observed band in bridging OH groups at about 3610 cm^–1 in H-mordenite is deconvoluted into two component bands, one at a higher frequency of 3612 cm^–1 and another at a lower frequency of 3585 cm^–1, which are assigned to OH groups in 12- and 8-ring channels, respectively [32-35]. The low-frequency-component IR band is attributed to Brӧnsted acid sites in a small channel, while the high-frequency one is attributed to those in a large ring, such as 8- or 10-rings [36-38].

Fig. 4(a) describes that pyridine adsorption in HZSM-35 zeolite takes place only on a part of the bridging OH groups in vacuum at room temperature with a small decrease in intensity at 3599 cm^–1, while the smaller acetonitrile molecule interacts with all bridging OH groups under the same conditions as the intensity at 3599 cm^–1 immediately decreases. This in turn proves the heterogeneous distribution of internal bridging OH groups in HZSM-35 zeolite. The accessibility of the internal Bronsted acidic bridging OH groups for pyridine molecules is studied with respect to temperature, as shown in Fig. 4(b). The intensity of the IR band at 3599 cm^–1 characteristic for strong acidic bridging OH groups clearly decreases after adsorption at 373 K. With an increase in the temperature of pyridine adsorption, more adsorption of pyridine on bridging OH groups of HZSM-35 zeolite takes place as follows from a decline in the intensity of the band at 3599 cm^–1. This means that the internal bridging OH groups are accessible to large pyridine molecules at high temperature [39, 40]. The existence of two bands at 3597 and 3589 cm^–1 is revealed after pyridine adsorption at different temperatures. The bridging OH groups located in the smaller channels are not so easily accessible for large pyridine molecules as those in the main channels. It seems reasonable to assign these two bands at 3597 and 3589 cm^–1 to the bridging OH groups located in cages and 8-ring channels, respectively, which supports our decomposition results shown in Fig. 3.

Zeolites, which are crystalline alumino-silicate microporous materials, possess interesting intrinsic acidic features, such as Lewis and Brӧnsted acid sites. Brӧnsted acid sites exist as bridging OH groups to Al and Si atoms, and their protons can be exchanged with other metal cations. Herein, we prepare ZSM-35 zeolites with varying concentrations of Brӧnsted acid through the sodium ion-exchange process. The crystalline structure and textural property are not affected by the exchange process, as shown in Fig. S1 and S2.

The IR characterization is used to clarify the change in the acidic properties of zeolites after the treatment. The intensity of the bridging OH groups at 3599 cm^–1 suffers a significant decrease after sodium ion exchange, as illustrated in Fig. 5(a), which indicates that the Brӧnsted acidic proton of HZSM-35 zeolite is substituted by Na⁺. A higher exchange degree leads to a lower intensity of the Brӧnsted acid hydroxyl groups. In HZSM-35 zeolite, the IR band of PyH⁺ appears at 1545 cm^–1 after adsorption of pyridine at 623 K for 30 min with the band of pyridine adsorbed on typical Lewis acid sites at 1445 cm^–1 (Fig. 5(b)) [41, 42]. The Lewis acid sites in HZSM-35 are assigned to the extra-framework aluminum species. The intensity of the band at 1545 cm^–1 drops significantly with the pretreatment of the sodium ion-exchange process, which indicates a decrease in the concentration of Brӧnsted acid in ZSM-35 zeolite. As shown in Table 1, the concentration of the Brӧnsted acid in HZSM-35 zeolite, obtained by the integration of the band at 1545 cm^–1, is 0.049 mmol/g, which decreases to 0.037 mmol/g in Na-Z-25 zeolite and even falls to 0.011 mmol/g in Na-Z-49 zeolite. The concentration of the Lewis acid remains constant at 0.004 mmol/g.

The effect of this sodium ion-exchange process on the bulk acidity of zeolite is studied using NH₃-TPD, as described in Fig. 6. Two ammonia desorption peaks are found at 465 and 709 K in HZSM-35 (Fig. 6(a)), which shows that two types of acid sites, including the weak and the strong ones, exist in the sample. The intensity of the latter peak suffers a decrease in Na-Z-25 zeolite, and even becomes a broad tail in Na-Z-49 zeolite. Note that the strength of the strong acid does not change in the sodium ion-exchange process for its peak temperature being maintained constant at 709 K. Taken together, the above results show that a large amount of strong Brӧnsted acid is substituted by the sodium ion, resulting in a loss of strong acidity in HZSM-35 zeolite.

The performance of ZSM-35 zeolites with different concentrations of Brӧnsted acid in the aldol condensation reaction of FA and MAc is evaluated in a continuous-flow fixed-bed reaction tube as illustrated in Fig. 7. Herein, dimethoxymethane (DMM) is employed as the source of FA. The mass spectrum detection result shown in Fig. S3 substantiates the facile decomposition of DMM to attain FA with 100% conversion in Fig. S4 over all ZSM-35 zeolites. MAc conversion changes slightly over all ZSM-35 zeolites, as shown in Fig. S4. However, the yield and S_MA+AA are remarkably enhanced parallel to the increase in the concentration of Brӧnsted acid in zeolites. Here, note that AA is mainly derived from the hydrolysis of MA and the ratio of MA to AA varies from 1 to 4. Given that the concentration of Lewis acid remains constant at 0.004 mmol/g for all zeolites, it is suggested that Brӧnsted acid is an active site for the aldol condensation reaction of DMM and MAc. The rate-determined step of aldol condensation reaction in the gaseous phase is reported to be the keto-enol tautomerization of MAc. The enol form of MAc reacts with FA to produce MA and AA. With a high concentration of Brӧnsted acid, the keto-enol tautomerization equilibrium might shift significantly to an enol counterpart, which consecutively reacts with FA to give a high S_MA+AA and yield.

The deposited coke amounts of deactivated ZSM-35 zeolites with different concentrations of Brӧnsted acid after aldol condensation reaction for 6 h are determined by TGA characterization. As shown in Fig. 8, the TGA results show that the amounts of deposited coke in HZSM-35, Na-Z-25, and Na-Z-49 are 8%, 5.5%, and 2.5%, respectively, among which HZSM-35 zeolite has the highest coke amount. Considering the difference in Brӧnsted acid concentration of the three catalysts, the lowest amount of coke within Na-Z-49 zeolite is ascribed to its low concentration of Brӧnsted acid. From the first derivative curve of TGA, the consumption temperatures of coke species within HZSM-35, Na-Z-25, and Na-Z-49 zeolites are 904, 860, and 819 K, respectively. HZSM-35 zeolite has the hardest coke species, which mainly consists of benzene, naphthalene, phenanthrene, and their methyl-substituted derivatives, as described in Fig. S5. The unsaturated product, MA and AA, might continue to perform the Diels-Alder ring-closing reaction on Brӧnsted acid sites to form aromatic species. On the other hand, the methanol-to-hydrocarbons reaction is unavoidable at such high reaction temperature of 623 K in the presence of a strong Brӧnsted acid and an amount of C₁–C₅ hydrocarbon product is detected in the initial period of the first 60 min. This reaction might also cause the deactivation of zeolites.

The distribution of Brӧnsted acid in HZSM-35 zeolite was systemically studied by IR experiments, and 51% of the total Brӧnsted acid was found to be located in the cage, while 23% and 26% were distributed in 10- and 8-ring channels, respectively. Brӧnsted acid was an active site for aldol condensation reaction of DMM and MAc for producing MA and AA. A high concentration of Brӧnsted acid was beneficial to the selectivity and yield of MA and AA. This study provided more information about the acid-catalyzed aldol condensation reaction. Moreover, the accurate position of the site of aldol condensation reaction in the framework, whether in a cage or in a 10- or 8-ring channel, as well as the effect of Lewis acid sites, need to be studied further.