Aldol condensation of acetaldehyde over Zr-β zeolites with tailored Lewis acidity and passivated Brønsted sites: toward environmentally benign crotonaldehyde synthesis

Abstract

In response to the challenges of equipment corrosion and environmental pollution associated with the homogeneous catalytic production of crotonaldehyde via acetaldehyde aldol condensation, this work focused on the synthesis of Zr-β zeolites with Lewis acidity. The crotonaldehyde selectivity was significantly improved by tailoring the acidic properties of the zeolites. A comparative study of three distinct Lewis acid sites identified isolated framework tetra-coordinated Zr sites as the most efficient catalytic centers, which were successfully constructed using hydrothermal synthesis and liquid-phase incorporation methods. Zr-β zeolites synthesized via liquid-phase incorporation demonstrated higher conversion and selectivity owing to larger pore size, greater total Lewis acidity and a higher proportion of weak and medium Lewis acid sites. These properties were further optimized by adjusting the precursor and solvent during the synthesis process. In situ DRIFTS analysis revealed that the Lewis acid sites activated the α-H of acetaldehyde, forming carbanion intermediates essential for the enolization and subsequent aldol condensation. The main by-product, methyl cyclopentenone, was found to originate from the Prins reaction of sorbaldehyde formed through excessive aldol condensation, which was mediated by Brønsted acid sites. To suppress this side reaction, the zeolites were modified with alkali cations (Na⁺, K⁺) to selectively passivate the Brønsted acid sites while enhancing the Lewis acidity significantly. This strategy effectively reduced by-product formation and ultimately achieved a crotonaldehyde selectivity of 94.7%.

---

1. Introduction

Crotonaldehyde, known as trans-2-butenal, is a key organic chemical used in the production of potassium sorbate, a safe food preservative in increasing demand. However, the current industrial route for crotonaldehyde synthesis via the aldol condensation of acetaldehyde predominantly relies on homogeneous catalysts, such as sodium hydroxide solutions, which lead to significant challenges including equipment corrosion and the generation of environmentally unfriendly saline wastewater requiring high treatment costs.^1,2 Therefore, the development of efficient and environmentally benign heterogeneous catalytic systems is of great importance.

Acetaldehyde aldol condensation is a fundamental C–C coupling reaction, where the activation of the acetaldehyde carbonyl group is essential. In acid-catalyzed systems, the carbonyl oxygen is protonated to form an enol intermediate, which then undergoes nucleophilic addition with another protonated acetaldehyde molecule. ZrO₂ catalysts demonstrate superior catalytic performance compared with MgO, La₂O₃, Sm₂O₃, and Nd₂O₃, which are commonly used in C–C coupling reactions, but ZrO₂ catalysts suffer from rapid deactivation.³ In contrast, in base-catalyzed systems, the α-H of acetaldehyde was activated to form a carbanion, which attacks the carbonyl carbon of another acetaldehyde molecule.⁴ While low-concentration alkali cations supported on SiO₂ can effectively promote the aldol condensation, stronger cations tended to induce coking, and higher loadings favored hydrogenation by-products.^5,6 On the other hand, heteroatom zeolites have shown promise in catalyzing C–C coupling reactions due to their tunable acidic properties, shape selectivity, and excellent structural stability.^7–11 As the pivotal step in the conversion of acetaldehyde to 1,3-butadiene, its aldol condensation is effectively promoted by Lewis acid sites in heteroatom zeolites (e.g., Ta, Zn, Nb, Mg, and Zr), where the catalytic performance is directly proportional to the Lewis acid content.^12–17 Consequently, heteroatom zeolites with tailored Lewis acidity are anticipated to offer enhanced performance in the aldol condensation of acetaldehyde to crotonaldehyde.

Several methods have been developed to prepare heteroatom zeolites, including hydrothermal synthesis, steam-assisted crystallization, and post-synthesis incorporation. Hydrothermal synthesis typically produces highly crystalline zeolites with controlled morphology and stable active sites, which is favorable for industrial applications, although it may lead to the formation of closed and less accessible sites.^18–22 Steam-assisted crystallization allows for the rapid synthesis of pure and well-ordered zeolites, but it is limited in the incorporation of heteroatoms into the framework.^23,24 Post-synthesis incorporation, particularly liquid-phase incorporation (LPI), offers an effective route for introducing heteroatoms into the parent zeolite. The dissolution and dispersion of metal precursors in solvents play a critical role in determining the framework metal content and the overall acidic properties, and thus represent a key factor in optimizing the catalytic performance of Zr-β zeolites.^25–28

The proposed reaction mechanism for the acetaldehyde aldol condensation to crotonaldehyde over Lewis acid sites of heteroatom zeolites may proceed as follows. Two acetaldehyde molecules co-adsorb on open tetra-coordinated metal sites, undergoing sequential proton transfer steps. The intermediate species are stabilized through M-OH species, which lower the activation energy and enhance catalytic performance.²⁹ Despite these insights, the detailed mechanism of this transformation remains to be fully elucidated and requires further investigation.

In this study, Zr-β zeolites with Lewis acidity were synthesized with the parent Hβ zeolite, aiming to catalyze the formation of crotonaldehyde from acetaldehyde. By comparing three different types of Lewis acid sites, the speciation of the most effective active site was identified. The catalytic performances of Zr-β zeolites prepared via hydrothermal synthesis and LPI were investigated to explore the structure–activity relationship between the acidic properties, pore properties and catalytic performance. Moreover, in situ diffuse reflectance infrared Fourier-transform spectroscopy (DRIFTS) was employed to probe the reaction mechanism. Based on the mechanistic insights, the modification with alkali cations (Na⁺, K⁺) was proposed to selectively passivate Brønsted acid sites while preserving the Lewis acidity, which suppressed the formation of by-products such as methyl cyclopentenone, leading to a crotonaldehyde selectivity of 94.7%.

2. Experimental

2.1. Catalyst preparation

Preparation of dealuminated Hβ zeolite (deAl-β, Si/Al = 1948). The commercial Hβ zeolite (Si/Al = 22) was stirred with nitric acid (Tianjin Jiangtian, 65%) in a ratio of 20 : 1 at 80 °C for 12 h, filtered, washed to neutrality, and dried at 110 °C for more than 6 h.

Preparation of ZrO₂/deAl-β. The ZrO₂/deAl-β sample was prepared by grinding the ZrO₂ solid and the deAl-β zeolite into powder and mixing thoroughly.

Preparation of Zr-β zeolite by hydrothermal synthesis method (HS-Zr-β). ZrCl₄ (Strem Chemicals, 99.5%) was mixed with tetraethylammonium hydroxide (TEAOH, Aladdin, 35%) under stirring at 60 °C for 1 h. After cooling to room temperature, gaseous SiO₂ (Aladdin, 99.8%) was added to deAl-β as seed crystal while stirring for 2 h. Then, NH₄F (Aladdin, AR) was added to obtain a gel molar composition of 1.0SiO₂ : 0.54TEAOH : 0.01ZrCl₄ : 0.54NH₄F : 8.2H₂O. The gel was dehydrated at 60 °C until the final H₂O/SiO₂ molar ratio was 3.9 to obtain a precursor gel with low water content. Subsequently, the gel was transferred to a high-pressure reaction vessel and crystallized at 180 °C for 24 h. When naturally cooled to room temperature, it was filtered and washed until neutrality. Finally, it was dried at 120 °C for 6 h and calcined at 550 °C in an air atmosphere for 6 h.

Preparation of Zr-β zeolite by LPI method (LPI-Zr-β). The precursor ZrCl₄ was dissolved in water while deAl-β (Si/Zr = 100) was added. The mixture was stirred at room temperature for 12 h and heated to 80 °C to evaporate the liquid. The obtained solid was ground into powder and calcined at 550 °C for 4 h in an air atmosphere. If the Zr-β zeolite was prepared with different precursors, such as ZrCl₄, ZrOCl₂·8H₂O (Guangfu, 99%) or Zr(NO₃)₄·5H₂O (Kemiou, AR), and other steps were kept the same as the LPI method, the obtained sample was denoted as Zr-β-M, where M is the type of precursor used. If the precursor ZrCl₄ was dissolved in different solvents, such as water (H₂O), ethanol (CH₃CH₂OH, Tianjin Jiangtian, AR), isopropanol ((CH₃)₂CHOH, Aladdin, AR) or dichloromethane (CH₂Cl₂, Aladdin, AR), and other steps were kept the same as the LPI method, the obtained sample was denoted as Zr-β-N, where N is the type of solvent used. If the precursor ZrCl₄ was dissolved in different solutions of alkali cation, such as NaNO₃ (Aladdin, AR) or KNO₃ (Aladdin, AR), and other steps were kept the same as the LPI method, the obtained sample was denoted as y-Zr-β, where y is the type of alkali cation.

2.2. Catalyst characterization

X-ray diffraction (XRD) patterns were obtained using an X-ray diffractometer (Rigaku Miniflex 600) under the conditions of Cu Kα radiation (λ = 0.154 nm), 40 kV, 15 mA, 5–50°, 20° min⁻¹. The relative crystallinity (RC) was defined as the ratio of its peak area at 2θ = 22.4° to that of the Hβ zeolite. Pore texture analysis was performed on an automatic physical adsorption instrument (Micromeritics Tristar 3000), and N₂ adsorption–desorption isotherms were measured to calculate BET surface area and pore volume after pretreatment (200 °C, 6 h vacuum). The morphology and particle size were observed by scanning electron microscopy (SEM) using a JSM-IT800 instrument. X-ray photoelectron spectroscopy (XPS) measurements were performed on a Thermo Scientific ESCALAB 250Xi spectrometer equipped with a monochromatic Al Kα X-ray source (1486.6 eV, 150 W) at a pass energy of 30 eV. Fourier-transform infrared (FT-IR) spectra were recorded on a Nicolet 6700 spectrometer (Thermo Fisher Scientific, USA) under the conditions of 400–4000 cm⁻¹, 4 cm⁻¹ resolution, and 64 scans. Ultraviolet-visible (UV-vis) diffuse reflectance spectra were recorded on a Perkin Elmer Lambda 750 spectrometer with a scanning range of 190–600 nm.

The acidic properties were analyzed by pyridine adsorption infrared (Py-IR) spectroscopy on an FTIR-650 spectrometer. Prior to adsorption, the sample was vacuum-dried at 350 °C for 1 h to remove impurities. After cooling to 50 °C, pyridine vapor (25 mbar) was introduced for 10 min to saturate acid sites, followed by 30 min vacuum evacuation to eliminate physisorbed pyridine. Temperature-programmed desorption was conducted by heating the sample to 150 °C, 250 °C, and 350 °C at a rate of 10 °C min⁻¹ under vacuum, with in situ infrared spectra collected after 30 min stabilization at each temperature. Acid site quantification utilized the integrated peak areas of Lewis (1450 cm⁻¹) and Brønsted (1540 cm⁻¹) acid bands, calculated via Lambert–Beer's law.³⁰ Total acidity was derived from pyridine adsorption at 150 °C, strong acidity from residual peaks at 350 °C, medium acidity from the difference between 250 °C and 350 °C peaks, and weak acidity from the difference between 150 °C and 250 °C peaks.

In situ DRIFTS was conducted as follows. The sample was placed in an in situ diffuse reflectance cell equipped with a water bath. The system was then purged with N₂ (50 mL min⁻¹) and pretreated at 220 °C for 1 h, followed by cooling to 200 °C. Background spectra were recorded prior to acetaldehyde introduction via N₂ bubbling. Time-resolved spectra (400–4000 cm⁻¹, 2 cm⁻¹ resolution, 64 scans) were collected at 2 min intervals with automatic background subtraction to monitor acetaldehyde adsorption. After 60 min, acetaldehyde feed was stopped, and N₂ purging was continued for another 60 min while spectra were similarly collected to track desorption behavior.

2.3. Catalyst evaluation

The reaction activity was evaluated in a fixed-bed reactor (8 mm inner diameter). The catalyst was pressed into pellets, sieved to 20–40 mesh, dried at 120 °C, and loaded (0.5 g) into the reaction tube. Then the system was purged with N₂ at 30 mL min⁻¹, heated to the specified temperature and stabilized for 1 h. An acetaldehyde aqueous solution was pumped into the reactor via a plunger pump. Samples were collected quantitatively using a six-port valve and analyzed online by a gas chromatograph equipped with an HP-PLOT/Q column (30 m × 0.32 mm × 10 μm) and an FID detector with He (30 mL min⁻¹) as the carrier gas and an injector temperature of 220 °C.

The reaction conversion and selectivity were calculated using the following equations:

where X is a product except for acetaldehyde.

3. Results and discussion

3.1. Structural characterization results

3.1.1. deAl-β, LPI-Zr-β and ZrO₂/deAl-β. As shown in the UV-vis spectra (Fig. 1a), the deAl-β sample exhibited a characteristic peak at approximately 223 nm, which was associated with the silanol nests formed following the removal of framework Al in Hβ zeolite. In contrast, the LPI-Zr-β sample displayed a distinct peak at around 197 nm, corresponding to isolated tetra-coordinated Zr species incorporated into the zeolite framework. The absence of the 223 nm peak in LPI-Zr-β suggested that Zr species had effectively interacted with the silanol nests, leading to the formation of Zr–O–Si bonds during the LPI process. For the ZrO₂/deAl-β catalyst, a peak at approximately 200 nm was observed, indicating the presence of highly dispersed ZrO₂ nanoparticles on the surface of the zeolite. Additionally, a new peak at around 229 nm was detected, which was attributed to the larger ZrO₂ particles.³¹

Fig. 1 (a) UV-vis and (b) Py-IR spectra of deAl-β, LPI-Zr-β and ZrO₂/deAl-β.

In the Py-IR spectra of the three samples (Fig. 1b), the peaks at 1445 cm⁻¹, 1490 cm⁻¹, and 1595 cm⁻¹ represented the characteristic peaks of pyridine molecules at Lewis acid sites. Additionally, the characteristic peaks at 1490 cm⁻¹ and 1541 cm⁻¹ were attributed to pyridine molecules adsorbed at Brønsted acid sites.^32,33 As summarized in Table S1, all three catalysts exhibited pronounced Lewis acidic property. The deAl-β sample, in which framework Al was removed, contained a large number of silanol nests, leading to a strong Lewis acidic property.³⁴ During the synthesis of the LPI-Zr-β sample, Zr species were incorporated into the silanol nests, forming isolated Zr sites with well-defined Lewis acidity. For the ZrO₂/deAl-β sample, the Lewis acid sites originated from both the extra-framework ZrO₂ species and the inherent Lewis acidity of the deAl-β support.

3.1.2. HS-Zr-β and LPI-Zr-β. In the XRD patterns of HS-Zr-β and LPI-Zr-β (Fig. 2a), the diffraction peaks at approximately 2θ = 7.8° and 22.4° were assigned to the (101) and (302) crystallographic planes of the *BEA topology in Hβ zeolite. Compared with the parent Hβ zeolite, HS-Zr-β displayed higher peak intensity and RC in these characteristic diffraction peaks. This improvement was likely attributed to the mineralizing effect of fluorine, which promoted the formation of a more ordered framework structure with fewer defects and enhanced morphological perfection.^35,36 In both Zr-β samples, the absence of diffraction peaks at 2θ = 28.2° and 31.5°, which were typically associated with the (−111) and (111) crystallographic planes of ZrO₂, suggested that the Zr species were highly dispersed within the zeolite framework and did not form large ZrO₂ particles.^37,38

Fig. 2 (a) XRD and (b) UV-vis spectra of HS-Zr-β and LPI-Zr-β.

The SEM images (Fig. S1) further revealed distinct morphological features between the two samples. HS-Zr-β exhibited an average particle size of 690 nm and a well-defined spherical morphology, while LPI-Zr-β was composed of smaller, aggregated particles with a size range of 100–200 nm. The pore properties of the two Zr-β zeolites are summarized in Table S2. Compared with HS-Zr-β, the total pore volume of LPI-Zr-β increased from 0.306 to 0.445 cm³ g⁻¹. It is speculated that fluoride in the hydrothermal synthesis resulted in larger and low-defect crystals with limited mesoporosity. Furthermore, high-temperature calcination to remove the template agent may block some pores, reducing the total pore volume.^39–41

In the UV-vis spectra (Fig. 2b), the characteristic absorption peak at 197 nm was attributed to the presence of isolated tetra-coordinated Zr species within the zeolite framework. The absence of a distinct peak at 230 nm in both samples suggested that no ZrO₂ species were formed, indicating that Zr was successfully incorporated into the framework and remained as isolated active sites during the synthesis process.⁴² As shown in Table S1, the Brønsted acidity of LPI-Zr-β was comparable to that of HS-Zr-β. However, LPI-Zr-β exhibited a significantly higher total Lewis acidity, with L/B approximately twice that of HS-Zr-β. Moreover, a higher proportion of weak and medium Lewis acid sites was observed in LPI-Zr-β.

3.1.3. Zr-β zeolites synthesized using different Zr precursors. In the XRD analysis of Zr-β zeolites synthesized using different Zr precursors (Fig. 3a), the diffraction peak corresponding to the (302) crystal plane of the deAl-β sample, which was obtained after dealumination of the Hβ zeolite, shifted to a higher diffraction angle, from 22.4° to 22.8°. This shift was likely attributed to the contraction of the zeolite unit cell induced by the removal of framework Al. After undergoing the LPI method, the diffraction peaks of the three Zr-β samples shifted to lower angles, approximately at 22.6°, suggesting that Zr atoms were successfully incorporated into the zeolite framework, resulting in an expansion of the unit cell. Moreover, the absence of characteristic diffraction peaks corresponding to ZrO₂ further confirmed the lack of ZrO₂ species in the three Zr-β zeolites.

Fig. 3 (a) XRD, (b) UV-vis, (c) Zr 3d XPS and (d) Py-IR spectra of Zr-β zeolites prepared with different precursors.

As shown in the SEM analysis (Fig. S2), the incorporation of Zr into the zeolite framework did not lead to noticeable morphological degradation or particle aggregation. All three Zr-β zeolites exhibited a similar morphology, consisting of particle clusters in the size range of 100–200 nm. The BET data are summarized in Table S2. There was only minor variation in the specific surface area and pore volume before and after dealumination, suggesting that the dealumination process caused minimal structural damage to the zeolite framework and retained a high crystallinity degree. Furthermore, the pore properties remained largely unchanged after Zr incorporation, indicating that the framework structure was well preserved and that the type of Zr precursor used had no significant influence on the pore architecture during the incorporation process.

In the UV-vis analysis (Fig. 3b), all three Zr-β zeolites displayed a distinct characteristic absorption peak at approximately 197 nm, which was attributed to the electron transfer between Zr species and O atoms within the zeolite framework. Similarly, the absence of a peak in the vicinity of 230 nm confirmed no extra-framework ZrO₂ species, indicating that Zr existed predominantly as isolated framework sites.

To further investigate the electronic states of Zr in the three Zr-β zeolites, XPS characterization was conducted (Fig. 3c). All samples exhibited characteristic Zr 3d peaks at 185.7 eV and 183.2 eV, corresponding to Zr 3d_3/2 and Zr 3d_5/2, respectively. In contrast, the binding energies for Zr 3d_3/2 and Zr 3d_5/2 in ZrO₂ were 184.6 eV and 182.2 eV, respectively. During zeolite synthesis, Zr atoms incorporated into the zeolite framework to form Zr–O–Si bonds. Since Si had a higher electronegativity than Zr, the electron cloud density around Zr atoms decreased, thereby shifting the Zr 3d binding energy to higher values.^43,44

The Py-IR spectra of the three Zr-β zeolites are presented in Fig. 3d, and the corresponding acidity data are summarized in Table S1. The strong absorption bands at 1445 cm⁻¹ and 1596 cm⁻¹ were attributed to the presence of Lewis acid sites in the zeolite framework, whereas the weaker band at 1542 cm⁻¹ was associated with Brønsted acid sites. Compared with the parent Hβ zeolite, the LPI-Zr-β zeolites exhibited a significant reduction in Brønsted acidity. In contrast, both the total Lewis acidity and the proportion of weak Lewis acid sites were notably increased. These results suggested that the LPI method effectively introduced Zr into the framework in a manner that enhanced the Lewis acidity while diminishing the Brønsted acidity.

3.1.4. Zr-β zeolites synthesized using different Zr solvents. Combining the XRD patterns (Fig. S3a) with the BET data summarized in Table S2, all samples displayed the characteristic diffraction peaks of the *BEA topology, with RC exceeding 70%. This suggested that the framework structure of the parent Hβ zeolite was well preserved during the synthesis process.

As observed in Fig. S4, no significant differences in morphology were detected among the four samples, all of which exhibited irregular aggregates composed of small particles with sizes in the range of 100–200 nm. Notably, the Zr-β zeolites prepared with organic solvents showed slightly higher RC, along with enhanced specific surface area and micropore volume, compared with Zr-β-H₂O. It was inferred that in organic solvents, ZrCl₄ was less prone to hydrolysis and can be more effectively dispersed within the zeolite channels, thereby minimizing pore blockage. Among the organic solvents, Zr-β-CH₂Cl₂ demonstrated a significant increase in both specific surface area and micropore volume relative to Zr-β-H₂O, indicating that ZrCl₄ was efficiently dispersed in CH₂Cl₂.

All four samples displayed distinct characteristic absorption peaks at 197 nm, corresponding to framework Zr species, with no observable absorption attributed to ZrO₂ species (Fig. S3b). This suggested that Zr was predominantly present as isolated framework sites within the zeolite structure. The XPS data (Fig. S3c) revealed that the binding energy of the Zr 3d_5/2 peak for all four Zr-β zeolites was approximately 183.3 eV, which was higher than the 182.2 eV typically observed for ZrO₂. This positive shift in binding energy indicated that Zr had been successfully incorporated into the zeolite framework, forming Zr–O–Si bonds.

In the Py-IR spectra (Fig. S3d), a prominent absorption peak at 1445 cm⁻¹ was assigned to Lewis acid sites, while a weaker peak at 1541 cm⁻¹ was associated with Brønsted acid sites. In line with the acidity data summarized in Table S1, the L/B for all samples was above 10, confirming the dominance of Lewis acidity in Zr-β zeolites. Compared with Zr-β-H₂O, the zeolites synthesized in organic solvents exhibited a significantly increased total amount of Lewis acid sites, although the proportion of strong Lewis acid sites was markedly reduced. Notably, Zr-β-CH₂Cl₂ showed the highest contribution of weak Lewis acid sites (56.0%) and the lowest level of strong Lewis acid sites (3.3%), suggesting that CH₂Cl₂ used as the solvent effectively modulated the acidity type and strength of the zeolite.

3.1.5. Zr-β zeolites modified with alkali cations. During the modification process, the alkali cations can undergo ion exchange with silanol groups in “open” Zr sites through hydrolysis of the Si–O–Zr bonds.⁴⁵ As shown in the XRD patterns (Fig. 4a), both modified samples retained the diffraction peaks corresponding to the *BEA topology, and no distinct diffraction features attributable to ZrO₂ species were observed. This suggested that the introduction of alkali cations (Na⁺ and K⁺) did not induce structural collapse of the zeolite framework or lead to the agglomeration of Zr species. The BET data summarized in Table S2 further supported this conclusion. The modification with alkali cations had a minimal impact on the pore properties of the zeolite, preserving the pore architecture and maintaining the RC. These findings confirmed that the zeolite framework remained intact and that the zeolite channels were not significantly blocked during the modification process.

Fig. 4 (a) XRD, (b) UV-vis, (c) XPS, (d) FT-IR and (e) Py-IR spectra of Zr-β zeolite before and after the modification.

UV-vis analysis (Fig. 4b) revealed that all three samples exhibited a distinct characteristic absorption peak at 197 nm, corresponding to the framework Zr species. No absorption peak attributable to ZrO₂ species was observed, indicating that the Zr species remained in the form of isolated framework sites after the modification. This result was consistent with the XRD data, which also suggested no structural disruption or phase segregation. In XPS spectra (Fig. 4c), the binding energy of the Zr 3d orbital was observed to shift to higher values. As shown in FT-IR spectra (Fig. 4d), a prominent peak at 950 cm⁻¹ in the deAl-β sample was assigned to the flexural vibration of Si–OH groups, indicating the formation of abundant silanol nests in the zeolite framework as a result of dealumination. After the modification with alkali cations, this 950 cm⁻¹ peak was no longer detectable, and a new band appeared at 962 cm⁻¹, which was ascribed to the Zr–O–Si bond between Zr and the zeolite framework.⁴⁶ This shift suggested that the modification process did not hinder the incorporation of Zr into the framework, and instead, it promoted the formation of Zr–O–Si species.

Based on the XPS results, it was inferred that the modification with alkali cations could effectively passivate the Brønsted acid sites. This hypothesis was further validated by Py-IR analysis (Fig. 4e). After the modification, the peak intensity at 1542 cm⁻¹, corresponding to pyridine adsorption on Brønsted acid sites, was significantly reduced. In contrast, the intensity of Lewis-acid-related peaks remained largely unchanged. As summarized in Table S1, the modification with alkali cations led to a marked decrease in Brønsted acidity, while the Lewis acidity was significantly enhanced. It is speculated that the introduction of alkali cations reduces the electron density around Zr atoms, enhancing their affinity for electron pairs. This further strengthens the Lewis acidity of Zr atoms, resulting in a significant increase in its Lewis acidity. Moreover, the strength distribution of the Lewis acid sites was not affected.

3.2. Reaction results

3.2.1. deAl-β, LPI-Zr-β and ZrO₂/deAl-β. The catalytic performance of the three samples was evaluated under identical reaction conditions (Fig. 5). Among them, LPI-Zr-β exhibited the highest conversion, reaching 22.8% after 6 h of reaction, along with the highest selectivity to crotonaldehyde (85.2%). In contrast, both deAl-β and ZrO₂/deAl-β showed significantly lower conversions, approximately 8% after the same reaction time. The crotonaldehyde selectivity of ZrO₂/deAl-β was around 80%, while that of deAl-β remained the lowest.

Fig. 5 Diagrams of (a) conversion and (b) crotonaldehyde selectivity over deAl-β, LPI-Zr-β and ZrO₂/deAl-β (reaction conditions: 20 wt% acetaldehyde aqueous solution, 200 °C, atmospheric pressure, WHSV = 0.25 h⁻¹, TOS = 6 h).

As presented in Table S3, the product distributions varied among the catalysts. For LPI-Zr-β, the main by-product was methyl cyclopentenone, with only minor contributions from other by-products such as 1,3-butadiene, ethanol, acetic acid, and ethyl acetate. The deAl-β sample also produced methyl cyclopentenone as the primary by-product. However, for the ZrO₂/deAl-β sample, acetic acid was the dominant by-product, followed by methyl cyclopentenone, and notable selectivity was observed toward ethanol and other recombination by-products.

These results suggested that the isolated Zr sites in the zeolite framework, as present in LPI-Zr-β, were the primary active sites capable of efficiently promoting the aldol condensation of acetaldehyde to crotonaldehyde. In contrast, the extra-framework ZrO₂ species in ZrO₂/deAl-β demonstrated inferior catalytic performance. This may demonstrate the mechanism by which isolated tetra-coordinated Zr species stably adsorb acetaldehyde molecules and their intermediates, whereas extra-framework ZrO₂ species exhibit weaker adsorption of acetaldehyde molecules.

3.2.2. HS-Zr-β and LPI-Zr-β. As shown in Table S3, while the conversions of the two samples were comparable, LPI-Zr-β exhibited a crotonaldehyde selectivity of 85.2%, which was approximately 6.5% higher than that of HS-Zr-β. This result indicated that LPI-Zr-β demonstrated superior catalytic performance for the aldol condensation of acetaldehyde, and methyl cyclopentenone was the main by-product for both samples. However, HS-Zr-β showed a higher selectivity toward this by-product compared with LPI-Zr-β. It was speculated that the higher Brønsted acidity or the presence of stronger Lewis acid sites in HS-Zr-β may promote excessive condensation and other side reactions, thus reducing the selectivity toward the desired product.

Furthermore, the HS-Zr-β zeolite exhibited a smaller pore size, which could hinder reactant access to the active sites and increase the diffusion resistance of the reaction products. This effect may lead to the accumulation of by-products within the pores. In contrast, the LPI method yielded a more favorable pore structure for mass transfer and product desorption, contributing to its enhanced catalytic selectivity. Therefore, the LPI method was selected as the preferred synthesis strategy for further investigations due to its superior catalytic performance in promoting the formation of crotonaldehyde from acetaldehyde.

3.2.3. Zr-β zeolites synthesized using different Zr precursors. The activity evaluation results (Fig. S5) indicated that the conversions of the Zr-β zeolites prepared with different Zr precursors were comparable (∼20%) under identical reaction conditions. However, the selectivity toward crotonaldehyde exhibited notable differences after 12 h of reaction. Specifically, Zr-β-ZrOCl₂ showed the lowest crotonaldehyde selectivity (63.2%), whereas Zr-β-ZrCl₄ demonstrated the highest selectivity (88.2%). In conjunction with the acidity data, it can be inferred that the superior selectivity of Zr-β-ZrCl₄ was attributed to its higher proportion of weak Lewis acid sites, which likely promoted the formation of crotonaldehyde over excessive condensation or other side reactions.

As summarized in Table S3, methyl cyclopentenone was identified as the main by-product for all three zeolites. Notably, the selectivity toward this by-product was the lowest for Zr-β-ZrCl₄ (6.5%) and the highest for Zr-β-ZrOCl₂ (17.4%). In addition, Zr-β-ZrOCl₂ exhibited relatively higher selectivity toward acetic acid and other recombination by-products, further suggesting a tendency toward undesired side reactions. Based on the above results, ZrCl₄ precursor was chosen for further studies, as the Zr-β-ZrCl₄ zeolite exhibited outstanding catalytic performance in favoring crotonaldehyde formation with minimal by-product generation.

3.2.4. Zr-β zeolites synthesized using different Zr solvents. The catalytic performance of Zr-β zeolites synthesized with four different solvents was evaluated (Fig. S6). Compared with those prepared in organic solvents, Zr-β-H₂O showed a slightly higher conversion. This was likely due to the higher solubility of ZrCl₄ in H₂O, which facilitated the formation of a greater number of active sites in the zeolite. Among these samples, Zr-β-CH₂Cl₂ exhibited the highest crotonaldehyde selectivity, reaching 93% after 12 h of reaction time. In conjunction with the acidity data from Table S1, it was inferred that weak Lewis acid sites played a key role in promoting the aldol condensation reaction, whereas strong Lewis acid sites favored side reactions, thereby reducing the selectivity toward the desired product.

As shown in Table S3, methyl cyclopentenone was the main by-product across all samples. Compared with the Zr-β-H₂O zeolite, those synthesized with organic solvents generally demonstrated lower by-product selectivity and improved crotonaldehyde selectivity. Notably, Zr-β-CH₂Cl₂ achieved the best selectivity due to its higher proportion of weak Lewis acidity and lower content of strong Lewis acid sites. Despite the high selectivity of Zr-β-CH₂Cl₂, its conversion remained relatively low. Furthermore, the low boiling point and high volatility of CH₂Cl₂ raised concerns regarding safety and environmental impact in industrial applications. Therefore, H₂O was ultimately chosen as the solvent for the preparation of Zr-β zeolites in the subsequent studies.

3.2.5. Zr-β zeolites modified with alkali cations. The catalytic performance of the Zr-β zeolites before and after the modification was evaluated (Fig. S7). Although the conversion decreased by 5%, the key target selectivity toward crotonaldehyde increased by 6.5%, reaching 94.7% after 12 h of reaction time. As presented in Table S3, the conversion and overall product distribution showed minimal variation following modification. However, the selectivity of the major by-product, methyl cyclopentenone, was reduced by approximately 4.5%. Additionally, the selectivity toward other recombination by-products was also markedly suppressed.

In conjunction with the preceding analysis of the acidic properties, these results confirmed that the modification with alkali cations selectively passivated the Brønsted acid sites, which were primarily responsible for side reactions, while retaining the intrinsic Lewis acidity of the catalyst. This modulation of acid sites not only suppressed the formation of undesired by-products but also enhanced the selectivity toward the target product, crotonaldehyde, to 94.7%.

3.3. Mechanism research results

The reaction pathway of aldol condensation catalyzed by Zr-β Lewis acid sites was explored by in situ DRIFTS experiments (Fig. 6a). The adsorption and reaction processes of acetaldehyde molecules on the surface of Zr-β zeolite were investigated at 200 °C. At the initial stage of acetaldehyde introduction, the spectral peaks were primarily located in the C=O region. The peaks at 1761 cm⁻¹ and 1744 cm⁻¹ corresponded to the C=O characteristic peaks of acetaldehyde, while the peaks at 1725 cm⁻¹ and 1645 cm⁻¹ were attributed to the C=O and C=C characteristic peaks of crotonaldehyde, respectively.⁴⁷ A weak peak at 1652 cm⁻¹ was presumed to originate from the acyl oxygen stretching vibration of the main by-product methyl cyclopentenone. These results suggested that acetaldehyde was rapidly converted into crotonaldehyde via aldol condensation in the initial stage, accompanied by the formation of methyl cyclopentenone. With more acetaldehyde molecules absorbed in Zr-β zeolite over time, the C–H stretching vibration peak of acetaldehyde at 2700 cm⁻¹ along with its carbonyl peaks at 1761 cm⁻¹ and 1744 cm⁻¹ progressively intensified. Meanwhile, the peaks of crotonaldehyde at 1725 cm⁻¹ and 1645 cm⁻¹ and the by-product methyl cyclopentenone at 1652 cm⁻¹ also increased. In the hydroxyl region, a peak at 3460 cm⁻¹ emerged, attributed to hydrogen-bonding interactions between acetaldehyde molecules and Zr-β zeolite. A negative peak gradually developed at 3724 cm⁻¹, reflecting ethanol adsorption on hydroxyl sites in the zeolite. The peak at 1371 cm⁻¹ was assigned to the in-plane bending vibration of the hydroxyl group of ethanol, indicating gradual formation of ethanol as a by-product. Additionally, as acetaldehyde adsorbed on active sites and reacted, a peak at 1125 cm⁻¹ grew progressively, correlating to carbanion species adsorbed on Zr-β zeolite.⁴⁸ This indicated that Lewis acid sites effectively activated the α-H of acetaldehyde molecules to form critical carbanion intermediates for acetaldehyde enolization and subsequent aldol condensation. With the increase of time, the peak intensity of carbanion species continued to rise, demonstrating their stable existence and accumulation on Zr-β zeolite.

Fig. 6 DRIFTS of (a) acetaldehyde adsorption on the surface of Zr-β zeolite and (b) samples after N₂ blowback.

The above in situ DRIFTS results confirmed the mechanism of crotonaldehyde formation from acetaldehyde aldol condensation catalyzed by Zr-β zeolite, as illustrated in Scheme 1. Acetaldehyde molecules were adsorbed on the framework Zr site, and under its action, α-H dissociated to form carbanion species, which could be stably adsorbed on the Zr-β active sites and promote the continuation of the aldol condensation reaction. Another acetaldehyde molecule was activated by Lewis acid at the Zr site, in which the C=O bond was activated and subsequently attacked by carbanion species, resulting in a nucleophilic addition reaction to form 3-hydroxybutyraldehyde, which was then dehydrated to form crotonaldehyde.

Scheme 1 The formation pathway of crotonaldehyde from acetaldehyde aldol condensation catalyzed by Zr-β zeolite.

After acetaldehyde feeding was stopped, N₂ was used for back blowing to obtain in situ DFIFTS spectra (Fig. 6b). The peaks at 1761 cm⁻¹ and 1744 cm⁻¹ were characteristic peaks of acetaldehyde, the peak at 1725 cm⁻¹ was a characteristic peak of crotonaldehyde, and the peak at 1652 cm⁻¹ was a characteristic peak of methyl cyclopentenone. The area of each characteristic peak was integrated, and the peak area ratios of crotonaldehyde/acetaldehyde and methyl cyclopentenone/crotonaldehyde were calculated (Fig. 7a). It can be found that the ratio of crotonaldehyde/acetaldehyde gradually decreased with time, and the ratio of methyl cyclopentenone/crotonaldehyde gradually increased, indicating that the production rate of crotonaldehyde was lower than the consumption rate. In addition, the production rate of methyl cyclopentenone gradually increased, suggesting that methyl cyclopentenone may originate from the further reaction of crotonaldehyde.

Fig. 7 (a) Trends of crotonaldehyde/acetaldehyde and methyl cyclopentenone/crotonaldehyde, (b) selectivity to crotonaldehyde and methyl cyclopentenone as a function of contact time.

Fig. 7b illustrates the relationship between the selectivity of crotonaldehyde and methyl cyclopentenone versus contact time. It was observed that with increasing contact time, the selectivity of crotonaldehyde decreased slowly, whereas the selectivity of the by-product methyl cyclopentenone increased gradually. When extrapolated to 0 contact time, the initial selectivity of crotonaldehyde was about 97%, indicating that crotonaldehyde was generated immediately after the raw material acetaldehyde came in contact with Zr-β catalyst; thus crotonaldehyde was the primary product of the acetaldehyde aldol condensation reaction.⁴⁹ As the initial selectivity of cyclopentenone approached 0, this indicated that methyl cyclopentenone was a secondary product, which was derived from the further reaction of intermediate species in the process of acetaldehyde aldol condensation rather than the direct activation of acetaldehyde.

Scheme 2 illustrates the formation pathway of the main by-product methyl cyclopentenone in the reaction. Methyl cyclopentone can be formed by intramolecular aldol condensation of sorbaldehyde (2,4-hexadienal), a C₆ carbonyl-containing substance, which was derived from the excessive aldol condensation of acetaldehyde.⁵⁰ It was speculated that sorbaldehyde would undergo intramolecular cyclization via the Prins reaction to produce methyl cyclopentenone.⁵¹ The Prins reaction proceeded via residual Brønsted acid sites in Zr-β zeolite, where the C=O group of sorbaldehyde formed a carbocation intermediate. This intermediate attacked the C=C group, driving cyclization, followed by deprotonation to yield methyl cyclopentenone.

Scheme 2 The formation pathway of methyl cyclopentenone and the possible reaction network for the aldol condensation of acetaldehyde.

Therefore, the possible reaction process of acetaldehyde aldol condensation catalyzed by Zr-β zeolite was as shown in Scheme 2. Crotonaldehyde, the main target product, was obtained by acetaldehyde aldol condensation reaction. Some acetaldehyde underwent Tischenko reaction to produce ethyl acetate, which was then hydrolyzed to acetic acid and ethanol.⁵² Furthermore, crotonaldehyde reacted further with acetaldehyde to form sorbaldehyde, which underwent the Prins reaction to generate methyl cyclopentenone. Excessive aldol condensation may lead to other recombination by-products.

4. Conclusions

In this study, a synergistic strategy combining the enhancement of weak and medium Lewis acidity and the passivation of Brønsted acid sites in Zr-β zeolites was proposed and demonstrated to significantly improve the selectivity toward crotonaldehyde. The framework tetra-coordinated Zr sites with Lewis acidity were identified as the most efficient active sites for the acetaldehyde aldol condensation, while extra-framework ZrO₂ species showed relatively poor catalytic performance. These efficient active sites were successfully constructed by both hydrothermal synthesis and LPI methods. Compared with HS-Zr-β, LPI-Zr-β zeolites exhibited higher conversion and crotonaldehyde selectivity owing to their greater total Lewis acidity and a higher proportion of weak and medium Lewis acid sites. Moreover, the larger pore size in LPI-Zr-β may facilitate the timely diffusion of products, thereby inhibiting further excessive condensation reactions and reducing by-product formation. Then, ZrCl₄ as the precursor and H₂O as the solvent were used for further optimization of the LPI synthesis. Notably, Zr-β-CH₂Cl₂ achieved the best selectivity due to its higher proportion of weak Lewis acid sites and lower content of strong Lewis acid sites. In situ DRIFTS analysis revealed that the Lewis acid sites facilitated the activation of the α-H in acetaldehyde, leading to the formation of carbanion intermediates, which were crucial for the enolization step and subsequent aldol condensation. The major by-product, methyl cyclopentenone, was found to originate from the Prins reaction of sorbaldehyde, which was formed through excessive aldol condensation, and the Prins reaction was mediated by Brønsted acid sites. To suppress this side reaction, the zeolites were modified with alkali cations (Na⁺ and K⁺), selectively passivating the Brønsted acid sites while enhancing the Lewis acidity significantly. This modification effectively suppressed by-product formation and achieved a crotonaldehyde selectivity of 94.7%.

Author contributions

Haoxi Jiang: conceptualization, methodology, resources. Qian Ran: formal analysis, investigation, data curation, writing – original draft, visualization. Yingying Zhao: validation, data curation. Guochao Yang: validation, supervision. Lingtao Wang: conceptualization, writing – review & editing, validation, project administration, supervision.

Conflicts of interest

There are no conflicts of interest to declare.

Data availability

The data supporting this article have been included as part of the supplementary information (SI). Supplementary information is available. See DOI: https://doi.org/10.1039/d5cy01009h.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (No. 21978211), the Natural Science Foundation of Tianjin (No. 21JCZDJC00520), the Natural Science Foundation of Ningbo (No. 2021 J008), and Tianjin University (2021XT-0008).

References

1. Z. Geng, L. Sheng and K. Zhang, Sep. Purif. Technol., 2024, 335, 126123.
2. H. Zheng, G. Xie and X. Wang, Acta Phys.-Chim. Sin., 2010, 26, 3273–3277.
3. W. Ji, Y. Chen and H. Kung, Appl. Catal., A, 1997, 161, 93–104.
4. C. L. Perrin and K. L. Chang, J. Org. Chem, 2016, 81, 5631–5635.
5. W. Ji and Y. Chen, Gaodeng Xuexiao Huaxue Xuebao, 1997, 18, 277–280.
6. L. Yang, C. Zhang and Y. Ren, Organometallics, 2025, 44, 410–417.
7. T. Yan, W. Dai and G. Wu, ACS Catal., 2018, 8, 2760–2773.
8. V. V. Ordomsky, V. L. Sushkevich and I. I. Ivanova, J. Mol. Catal. A: Chem., 2010, 333, 85–93.
9. X. Si, M. Wu and X. Qiao, Fuel, 2025, 381, 133347.
10. Y. Chen, T. Lu and X. Yang, Mol. Catal., 2024, 559, 114093.
11. W. Kagunya and W. Jones, Appl. Clay Sci., 1995, 10, 95–102.
12. H. Jiang, L. Yi and G. Yang, Catal. Sci. Technol., 2023, 13, 5902–5911.
13. H. Jiang, X. Ni and A. Jiang, Catal. Commun., 2023, 183, 106777.
14. M. Zhang, Y. Zhang and Y. Tan, J. Chem. Technol. Biotechnol., 2023, 98, 1620–1630.
15. H. Jiang, Y. Gao and L. Yi, Chem. Eng. J., 2024, 499, 156298.
16. H. Jiang, L. Yi and G. Yang, Catal. Sci. Technol., 2023, 13, 245–256.
17. H. Gu, G. Yang and L. Wang, Mol. Catal., 2025, 570, 114666.
18. Y. Yue, G. Zhu and M. Liu, Catalysts, 2024, 14, 924–924.
19. F. Clark, R. Garcia and A. Turrina, Microporous Mesoporous Mater., 2024, 376, 113177.
20. M. A. Camblor, A. Corma and S. Valencia, J. Mater. Chem., 1998, 8, 2137–2145.
21. T. Blasco, M. A. Camblor and A. Corma, J. Phys. Chem. B, 1998, 102, 299–304.
22. C. S. Cundy and P. A. Cox, Microporous Mesoporous Mater., 2005, 82, 1–78.
23. J. Weitkamp and M. Hunger, Stud. Surf. Sci. Catal., 2005, 155, 1–12.
24. N. Jappar, Q. Xia and T. Tatsumi, J. Catal., 1998, 180, 132–141.
25. H. Li, S. Zhao and W. Zhang, Fuel, 2023, 342, 127786.
26. F. Guan, T. Ma and X. Yuan, Catal. Lett., 2018, 148, 443–453.
27. J. Wang, K. Okumura and S. Jaenicke, Appl. Catal., A, 2015, 493, 112–120.
28. Y. Chai, L. Xie and Z. Yu, Microporous Mesoporous Mater., 2018, 264, 230–239.
29. D. Palagin, V. L. Sushkevich and I. I. Ivanova, J. Phys. Chem. C, 2016, 120, 23566–23575.
30. C. Emeis, J. Catal., 1993, 141, 347–354.
31. E. M. Albuquerque, P. N. Paulino and R. Sadek, Appl. Catal., A, 2020, 604, 117766.
32. R. Buzzoni, S. Bordiga and G. Ricchiardi, Langmuir, 1996, 12, 930–940.
33. P. Li, G. Liu and H. Wu, J. Phys. Chem. C, 2011, 115, 3663–3670.
34. F. Yi, Y. Chen and Z. Tao, J. Catal., 2019, 380, 204–214.
35. T. Blasco, M. Camblor and A. Corma, J. Phys. Chem. B, 1998, 102, 75–88.
36. M. Zhu, Y. Liu and J. Ma, J. Sol-Gel Sci. Technol., 2025, 114, 1–11.
37. F. Gonell, M. Boronat and A. Corma, Catal. Sci. Technol., 2017, 7, 2865–2873.
38. S. Song, L. Di and G. Wu, Appl. Catal., B, 2017, 205, 393–403.
39. C. S. Cundy and P. A. Cox, Chem. Rev., 2003, 103, 663–701.
40. J. Dijkmans, J. Demol and K. Houthoofd, J. Catal., 2015, 330, 545–557.
41. J. Wang, K. Okumura and S. Jaenicke, Appl. Catal., A, 2015, 493, 112–120.
42. W. Li, S. Liu and H. Wang, Catal. Commun., 2020, 133, 105845–105850.
43. G. Yang, L. Wang and H. Jiang, React. Chem. Eng., 2020, 5, 1833–1844.
44. X. Li, J. Pang and C. Wang, Green Chem., 2020, 22, 2852–2861.
45. R. Otomo, R. Kosugi and Y. Kamiya, Catal. Sci. Technol., 2016, 6, 2787–2795.
46. B. Tang, W. Dai and X. Sun, Green Chem., 2015, 17, 1744–1755.
47. O. V. Larina, I. M. Remezovskyi and P. I. Kyriienko, React. Kinet., Mech. Catal., 2019, 127, 903–915.
48. A. Chieregato, J. Velasquezochoa and C. Bandinelli, ChemSusChem, 2015, 8, 377–388.
49. V. L. Sushkevich, I. I. Ivanova and V. V. Ordomsky, ChemSusChem, 2014, 7, 2527–2536.
50. R. Wang, Y. Liu and G. Li, ACS Catal., 2021, 11, 4810–4820.
51. E. S. Vasiliadou, S. Li and S. Caratzoulas, Catal. Sci. Technol., 2018, 8, 5794–5806.
52. S. Colley, J. Tabatabaei and K. Waugh, J. Catal., 2005, 236, 21–33.

---