Electronically tailored metal-ion-chelation strategy promotes ionic liquid catalysis at near-ambient condition

Abstract

Electronic properties of active sites profoundly influence catalytic production of value-added chemicals; however, rational description and modulation is still a significant challenge. Herein, we propose an effective metal-ion-chelation strategy guided by density functional theory prediction and in situ Raman observation to structurally tailor and quantitatively correlate the electronic properties of active sites in ionic liquid. Comprehensive characterizations and theoretical calculations, in combination with electronic properties–performance correlations, reveal the electronic peculiarity of nitrogen and oxygen centers can be controllably restructured for remarkable improvement of catalytic performance at near-ambient condition. The turnover frequency is increased by two folds with deactivation rate suppressed by more than one half, while the kilogram-scaled recycling pilot achieves similar performance for the probe methacrolein synthesis. This strategy further exhibits excellent applicability and tolerance in other substrates with representative functional groups. Our work expresses the significance of electronic properties and provides a valid regulation approach for ionic liquid catalysis.

Accurately describing and tuning active sites remains a major challenge. Here, the authors introduce an effective metal-ion chelation strategy—guided by DFT predictions and in situ Raman measurements—to structurally engineer and quantitatively link the electronic properties of active sites in an ionic liquid.

Introduction

The development of well-designed functional catalysts significantly facilitates the sustainable conversion of petroleum and coal-based downstream products into value-added chemicals^1–3, while the electronic properties of active sites directly affect the catalytic performance⁴. In contrast to heterogeneous catalytic materials, the description and controllable regulation of electronic properties of homogeneous molecular catalysts is more aligned with intrinsic principles, yet it poses considerable challenges in structural design^5,6. Ionic liquids (ILs) as a representative of green medium and catalyst have been extensively utilized in various environmentally friendly and mild catalytic processes, due to the flexible design of their cation and anion structures^7–13. During the transformation in IL catalysis system, the efficient activation of reactants through the nucleophilic or electrophilic interaction is dramatically influenced by the electronic properties of active centers (electron-rich or electron-deficient)^14–16. Unfortunately, the unbalanced electron density of catalytic sites often leads to a trade-off between the catalytic performance and stability, thereby limiting the efficiency of transformative process^17,18. Therefore, the rational description and fine-tuning of the electronic properties of active sites are essential to promote IL catalysis, especially in the aldol reaction of aldehydes¹⁹.

In recent years, the synthesis of value-added α, β-unsaturated aldehydes via direct aldol condensation between aldehydes has received considerable attention^20–22. In particular, the synthesis of methacrolein (MAL) from formaldehyde and propionaldehyde is a crucial step in the production of methyl methacrylate (MMA), which is a significantly important chemical material in the fine chemical, agrochemical, and pharmaceutical industries^23,24. Compared to the traditional sec-amine catalysis process at 100–210 °C and 4.5–6.0 MPa²⁵, the ILs show enhanced catalytic performance under near-ambient condition (25–50 °C and 0.1 MPa)^26–29. Nevertheless, the protonation of sec-amine center in IL leads to the difficulty in dissociation of this catalytically active sites and dramatically reduces the electron density of the nitrogen atom, which decreases the catalytic activity. Additionally, the undesired cyclization generates oxazolidine-type byproducts, which results in catalyst deactivation (Supplementary Fig. 1). Therefore, an effective strategy is urgently required to motivate catalytic activity by increasing the electron density of the nitrogen atom and suppress catalyst deactivation by stabilizing hydroxyl groups as well.

The potential factors influencing the electronic properties of active site in IL are complicated, primarily driven by the ionic bond between the cation and anion, along with polarization, inductive effect, hydrogen bonding, and van der Waals forces³⁰. Over the past decades, great efforts have been dedicated to tune the electronic properties. On one hand, series of multi-scale regulation methods have been developed, including nanoconfinement^31,32, multicomponent mixing³³, interface enhancement^34,35, and electric field assistance^36,37. On the other hand, progressive achievements in theoretical calculations^38–41 and multiple characterization techniques^42,43 enable researchers to explore the electronic properties and regulatory mechanisms in microscopic level. These investigations illuminate the critical role of electronic properties in determination of catalytic performance. Nonetheless, the relationship between the structure and electronic properties, as well as their underlying correlations to catalytic performance has not been fully elucidated, due to the inherent complexity of intermolecular interaction and local environment^44,45. Thus, it is still challenging to precisely tailor the electronic properties by structure regulation to promote catalytic performance.

Herein, we propose a novel metal-ion-chelation strategy to controllably tailor the electronic peculiarity of catalytically active site in IL with the guidance of theoretical calculations and in situ Raman observations, which significantly prompt catalytic activity and stability at near-ambient condition. The relationship between the electronic properties and performance derived from the descriptors identified the optimal catalyst among the synthesized metal-ion-chelated ILs (MILs). Furthermore, the comprehensive characterizations and density functional theory (DFT) calculations reveal the intramolecular induction and coordination effects resulting from metal-ion chelation restructure the nanoscale morphology and electronic peculiarity, which promotes the dissociation and stability of catalytically active sec-amine center. Consequentially, both catalytic activity and stability in probe aldol condensation can be significantly improved by structurally tailored electronic properties. This strategy also demonstrates remarkable application in other substrates with representative functional groups.

Results

Electronically tailored design of MIL

To modulate the electronic characteristic of the nitrogen atom in protic diethanolamine (DEA)-type IL utilizing the metal-ion-chelation strategy (Fig. 1a), we predicted the Hirshfeld charge of the central nitrogen atom, namely the catalytically active amine site, of various MILs with optimized structure using DFT method (Fig. 1b and Supplementary Tables 1–2). We first confirmed the negative effect of protonation on the Hirshfeld charge of the central nitrogen atom in protic DEA-type IL using [HDEA]Pc as an example. The [HDEA] cation features symmetrical linear ethanol groups with proton partially delocalized between the cation and anion, resulting in the formation of a N⁺–H···Pc⁻ group. As a result, the Hirshfeld charge on the central nitrogen atom is reduced from −0.1479 to −0.0772 after protonation of DEA. When the [HDEA]Pc IL is chelated with monovalent metal ions of Cs⁺, K⁺, Na⁺, and Li⁺ through the interaction with hydroxyl groups, a certain degree of bending in the linear structure of cation is caused. While minimal influence is exerted on the anion, thus only slight alteration is discovered on the Hirshfeld charge of the central nitrogen atom. As for the chelation with divalent metal ions of Ba²⁺ and Mg²⁺, the Hirshfeld charge of the central nitrogen atom rises to −0.0841 and −0.1031, respectively. Meanwhile, a more significant bending of the alkyl chain is noticed in cation part, suggesting the increasing electronegativity of divalent metal ions correlates the enhanced chelation with hydroxyl groups. Furthermore, a higher coordination number enables them to exhibit interaction with the carbonyl oxygen of anion. Conversely, when we examine the chelation of Mn²⁺ and Y³⁺, the Hirshfeld charge of the central nitrogen atom significantly decreases to −0.0789 and −0.0709, even lower than that of [HDEA]Pc IL. As a result, the MgIL with the highest Hirshfeld charge of the central nitrogen atom is identified as the optimal catalyst candidate for the aldol condensation of aldehydes.

Fig. 1. Rational design of MIL inspired by theoretical prediction and in situ observations.

a Schematic illustration of electronically tailored MIL. X⁻ represents CH₃CH₂COO⁻. b Theoretically calculated Hirshfeld charges of the central nitrogen atom in MIL models (M/IL = 1/1) along with their optimized structures. Atoms are represented in cyan, red, blue, and gray for C, O, N, and H, respectively, with the central metal ions labeled below the structures. c Relationships between the TOF as well as ln r_d, and Hirshfeld charge of the central nitrogen and hydroxyl oxygen atoms. The error bars represent the standard deviation of potential inconsistencies in the Hirshfeld charge of two different hydroxyl oxygen atoms. d Relationships between the ΔE_M–IL and both TOF and C_MAL. C_MAL is defined by the MAL concentration measured at the reaction time of 60 min. e Relationships between the ΔE_M–IL and both ln r_d and stability. The stability is defined as the percentage of remaining catalytically active amine species after 60 min of reaction.

We postulated that catalytic activity is closely related to the Hirshfeld charge of the central nitrogen atom, while catalytic stability depends on the charge of the hydroxyl oxygen atoms in the respective [HDEA]Pc IL and MIL. To confirm this hypothesis, the MIL catalyst series was first synthesized using the fundamental [HDEA]Pc IL. We then employed an in situ Raman spectroscopy system to quantitatively analyze the probe aldol reaction between the formaldehyde and propionaldehyde to produce MAL with the catalysis of [HDEA]Pc IL and MIL (Supplementary Figs. 2–14). Figure 1c indicates that the turnover frequency (TOF) value is positively correlated with the Hirshfeld charge of the central nitrogen atom in the corresponding MIL. Meanwhile, we found a linear relationship between the catalyst deactivation rate and Hirshfeld charge of the hydroxyl oxygen atoms in MIL. Therefore, it can be concluded that the Hirshfeld charges of the central nitrogen atom and hydroxyl oxygen atoms are strongly relevant to the catalytic activity and stability. To further investigate the effect of chelation by different metal-ions, the relationship between the interaction energy (ΔE_M–IL) and catalytic performance of MIL was established (Supplementary Table 3). Specifically, the lower ΔE_M–IL reflects the stronger interaction between the metal-ion and [HDEA]Pc IL⁴⁶. Figure 1d reveals that both TOF and C_MAL change in a volcano-shaped trend as the ΔE_M–IL increasing progressively from −29.2 to −131.6 kcal·mol⁻¹, with the peak value appearing at the position of Mg²⁺. The catalytic activity of YIL is even lower than that of pristine [HDEA]Pc IL, attributed to the strongest interaction and lowest Hirshfeld charge of the central nitrogen atom. Furthermore, the catalytic stability of MIL is observably improved in proportion with the decreased ΔE_M–IL, namely the catalyst deactivation can be inhibited by modulation of their interactions (Fig. 1e). These observations elucidate that the suitable interaction between the metal-ion and [HDEA]Pc IL via chelation will contribute to the relatively high Hirshfeld charge of the central nitrogen atom and thus electron-rich state of catalytically active amine site through the intramolecular inductive and coordination effects. Besides, the interaction between the metal-ion and hydroxyl groups also hinders the intramolecular cyclization of IMI intermediate. Thereby, the catalytic activity and stability in MAL production from aldol reaction of formaldehyde and propionaldehyde can be promoted. Nevertheless, excessively strong interactions will lead to a decrease in the Hirshfeld charge of the central nitrogen atom, which significantly diminishes catalytic activity despite the catalyst deactivation can be further suppressed. Accordingly, the theoretically predicted MgIL indeed exhibits high efficiency and excellent stability on the probe MAL synthesis via aldol reaction.

Characterization and controllable modulation of MgIL

With the optimal MgIL in hand, which is structurally confirmed using NMR (Supplementary Fig. 15), we further investigated and modulated the underlying physicochemical properties based on the electronic characteristics affected by the metal-ion chelation through changing the Mg²⁺ content. The cryogenic transmission electron microscope (Cryo-TEM) was employed to directly observe the morphological transformation of [HDEA]Pc IL after chelation with Mg²⁺. As the Mg²⁺ content increases, the intrinsic network structure of [HDEA]Pc IL gradually changes into spheroids (MgIL-1.0), with the metastable worm-like structure (MgIL-0.5) as transition state (Fig. 2a and Supplementary Fig. 16). Meanwhile, the average cluster size gradually decreases from 287.5 ± 2.9 ([HDEA]Pc IL) to 140.3 ± 0.7 nm (MgIL-0.5), and then to 93.8 ± 0.7 nm (MgIL-1.0), as determined by DLS (Fig. 2b). These observations at the nanoscale level demonstrate that the chelation of [HDEA]Pc IL with Mg²⁺ facilitates the formation of more shaped and uniformed microstructural morphology of MgIL. To analyze the electronic characteristics of Mg²⁺, X-ray photoelectron spectroscopy (XPS) was employed to measure and compare the binding energies of MgIL specimens having different Mg²⁺ content. As presented in Fig. 2c, the Mg 1s spectra illustrate that the peak located at 1304.0–1303.8 eV is consistent with the Mg²⁺ species reported in literature⁴⁷. Moreover, this peak shifts to lower energy compared to that of Mg(Ac)₂ (1304.4 eV) as the Mg²⁺ content increases, indicating that the enhanced chelation effect will give rise to the electron-rich state of Mg²⁺. Also, the MgIL-1.0 displayed lower white line intensity in X-ray absorption near-edge structure (XANES) spectrum, compared with that of Mg(Ac)₂, which further reveals the electron transfer from [HDEA]Pc IL to Mg²⁺ (Fig. 2d)^48,49. We also conducted ¹³C NMR experiments to test the chelation effect of Mg²⁺ on the hydroxyl groups of [HDEA], as the chemical shift of α-C atom reflects the changes of chemical environment of the adjacent oxygen atom. As displayed in Fig. 2e, we found an upfield chemical shift for [HDEA]Pc IL compared to DEA, while a series of successive downfield shifts for MgIL with increasing Mg²⁺ content. When the DEA is protonated by propionic acid (HPc), the decreased inductive effect from the central nitrogen atom increases the electron density of the α-C atom, which contributes to the electronic shielding and decrease in resonating frequency. In contrast, the increased resonating frequency indicates the deshielding effect on the carbon nucleus, explained by the electron donation from the hydroxyl oxygen atoms to Mg²⁺ during chelation, which reduces the electron density of α-C atom. These results demonstrate that the chelation effect will be enhanced with the increasing Mg²⁺ content, which is in alignment with the results obtained from the XPS Mg 1s spectra. In addition, the ¹⁵N NMR characterization was performed to catch insight into the effect of Mg²⁺ chelation on the electron density of the central nitrogen atom. It has been known that the isotropic nuclear magnetic deshielding of nitrogen atom will constantly shift downfield when it participates in the formation of an ionic bond^50,51. As illustration in Fig. 2f, we first notice a downfield chemical shift after the protonation of DEA by HPc, which is associated with the reduced electron density of the central nitrogen atom by ionization. However, subsequent upfield chemical shifts are found after the chelation of [HDEA]Pc IL with Mg²⁺, implying the increased electron-rich state of the central nitrogen atom with higher Mg²⁺ content in MgIL. In addition, these findings are further supported by the observed binding energy shift in XPS N 1s spectra of [HDEA]Pc IL after chelation with different content of Mg²⁺ (Supplementary Fig. 17). Furthermore, we synthesized a series of MgILs with different anions, including Cl⁻, NO₃⁻, SO₄²⁻ and CF₃SO₃⁻ to investigate the role of Ac⁻ introduced by Mg(Ac)₂. Supplementary Fig. 18 demonstrates the largest upfield ¹⁵N chemical shift when Ac⁻ introduced by Mg salts, suggesting the most effective to facilitate the electron-rich state of the nitrogen atom. Additionally, the combination of [HDEA]Ac and Mg(Ac)₂ shows a nearly identical ¹⁵N chemical shift to that of [HDEA]Pc and Mg(Ac)₂ system. These results suggest that the effect of Ac⁻ and Pc⁻ on the electronic properties of the nitrogen atom is almost similar when acting as anions in these designed MILs, but Ac⁻ shows more beneficial than Cl⁻, NO₃⁻, SO₄²⁻ or CF₃SO₃⁻.

Fig. 2. Morphological and electronic characterizations of MgIL.

a Cryo-TEM images of [HDEA]Pc IL, MgIL-0.5 and MgIL-1.0. b Cluster size distribution of [HDEA]Pc IL, MgIL-0.5, and MgIL-1.0. The error bars for the mean values represent the standard deviations calculated from three independent DLS measurements. c XPS Mg 1s spectra of Mg(Ac)₂ and MgIL-X (X = 0.25, 0.5, 0.75, 1.0). The samples are prepared using pristine [HDEA]Pc IL without water to avoid volatile components during XPS measurements. d XANES spectra of Mg(Ac)₂ and MgIL-1.0. e ¹³C NMR spectra of DEA, [HDEA]Pc IL, and MgIL-X (X = 0.25, 0.5, 0.75, 1.0). f ¹⁵N NMR spectra of DEA, [HDEA]Pc IL, and MgIL-X (X = 0.25, 0.5, 0.75, 1.0).

Aside from the intramolecular inductive and coordination effects derived from the chelation with Mg²⁺ that will trigger the increased electron density of the central nitrogen atom (namely amine site), the electrostatic interaction between the cation and anion and their local environment also plays an important role in electronic properties^52–54. We first performed the ionic conductivity measurements and isothermal titration calorimetry (ITC) analysis to explore the bindings between the Mg(Ac)₂ and [HDEA]Pc IL. It is held that the conductivity activation energy (E_a) measures the activation barrier of ion migration, while the ionic conductivity (σ) depends on both ion migration and concentration⁵⁵. As illustrated in Fig. 3a, both saturated aqueous Mg(Ac)₂ and [HDEA]Pc IL solutions exhibit relatively high σ but low E_k of 7.1 and 10.0 kJ·mol⁻¹, respectively, due to their intrinsically high ionization and ion migration capabilities. Oppositely, significantly lower σ and higher E_a are observed in the case of MgIL. Specifically, the σ declines substantially while E_k increases from 15.0 to 27.1 kJ·mol⁻¹ when the molar ratio of Mg²⁺ to [HDEA]Pc IL ranges from 0.25 to 1.0. It suggests that the bindings between the Mg(Ac)₂ and [HDEA]Pc IL are continuously strengthened with the rising Mg²⁺ content, which effectively restricts the migration of free ions. In other words, the chelation of [HDEA]Pc IL with Mg(Ac)₂ indeed occurs in their mixtures. The ITC investigations further reveal that the interaction between the Mg²⁺ and [HDEA]Pc IL is associated with a moderate endothermic process, involving approximate three binding sites (Fig. 3b). These findings signify that these three binding sites include two hydroxyl groups of [HDEA] cation and the other carbonyl group of Pc⁻ anion, which thereby support the validity of our proposed chelation structure for DFT calculation.

Fig. 3. Effect of Mg²⁺ chelation on the local environment of MgIL.

a Ionic conductivity test of saturated aqueous Mg(Ac)₂ solution, [HDEA]Pc IL, and MgIL-X (X = 0.25, 0.5, 0.75, 1.0) samples. b ITC investigations including the thermograms (upper plots) and binding isotherms (lower plots) for the chelation interaction between the Mg²⁺ and [HDEA]Pc IL. The binding isotherms are calculated using aqueous Mg(Ac)₂ solution (20 mmol∙L⁻¹) titrated into water as the blank control and fitted with a first-order binding model. c FT-IR spectra of Mg(Ac)₂, [HDEA]Pc IL, and MgIL-X (X = 0.25, 0.5, 0.75, 1.0) samples. d ¹H NMR spectra of DEA, [HDEA]Pc IL, and MgIL-X (X = 0.25, 0.5, 0.75, 1.0). e ATR-FIR spectra of [HDEA]Pc IL and MgIL-X (X = 0.25, 0.5, 0.75, 1.0) samples.

To further understand and modulate the impact of Mg²⁺ chelation on the interaction between the cation and anion, we employed Fourier transform infrared (FT-IR), ¹H NMR, and far-infrared (FIR) spectroscopies to monitor the evolution of molecular-level interactions as a function of Mg²⁺ concentration. The FT-IR spectroscopic measurements reveal a gradual blue shift from 1562 ([HDEA]Pc IL) to 1572 cm⁻¹ (MgIL-1.0) for the antisymmetric stretching vibration frequency of carbonyl group attributed to the Pc⁻ anion, which is distinguished from that in Ac⁻ at 1591 cm⁻¹ (Fig. 3c). It demonstrates the successively enhanced coordination effect of Mg²⁺ on the Pc⁻ anion with increasing Mg²⁺ content in MgIL. While the ¹H NMR spectra provide insights into the changes in hydrogen-bonding strength between the cation and anion (Fig. 3d). The downfield ¹H chemical shifts of these two pairs of methylene hydrogens in [HDEA]Pc IL suggest that the [HDEA] shows intense hydrogen-bonding interaction with Pc⁻ after protonation of DEA by HPc⁵⁶. However, we subsequently detected a series of upfield ¹H chemical shifts for the MgIL samples having different Mg²⁺ content, indicating that the hydrogen-bonding interaction between the cation and anion will be weakened as the Mg²⁺ content is raised. According to the reported literature, it is believed that the stronger (or weaker) interaction between the DEA and HPc through protonation will lead to greater shortening of the intermolecular N⁺–H bond, which is reflected by a blue (or red) shift in FIR stretching^57,58. As the content of Mg²⁺ in MgIL increases, the observed red shift of stretching vibration band identifies that the chelation effect of Mg²⁺ weakens the N⁺–H bond, which is benefit for the dissociation of sec-amine site for aldehyde activation (Fig. 3e). Thus, the chelation of Mg²⁺ also plays a pivotal role in modulating the local environment of cation and anion, ultimately in favor of fine-tuning the electronic properties of MgIL.

Catalytic performance and mechanism

Building on the impressive catalytic efficiency of MgIL, we embarked on a detailed investigation into the effect of Mg²⁺ content on the performance of MgIL. In situ Raman spectroscopy served as a powerful tool to monitor the dynamic evolution of catalytically active species consumption and formation of MAL from the probe aldol reaction of formaldehyde with propionaldehyde (Fig. 4a, b, and Supplementary Figs. 19–22). As the content of Mg²⁺ incrementally increased from 0 to 3.8 wt.% (corresponding to the molar ratio of Mg²⁺ to [HDEA]Pc IL from 0 to 1), we can observe a notable enhancement of TOF from 0.67 to 1.36 h⁻¹, while the C_MAL rises from 0.25 to 0.54 mmol·g⁻¹. These are accompanied by a dramatic reduction of r_d to approximately one-tenth, alongside a 33% enhancement in stability when contrasted with [HDEA]Pc IL (Fig. 4c). Our comparative analysis reveals a striking correlation between the higher Mg²⁺ content and improved catalytic performance, with MgIL-1.0 (featuring equimolar Mg²⁺ and [HDEA]Pc IL) emerging as the champion in the view of catalytic activity and stability (Fig. 4d). Additionally, we investigated the situation that Mg²⁺/[HDEA]Pc IL ratio exceeds 1.0, for instance 1.25, the catalyst becomes unstable, resulting in the precipitation of a white solid. Noting that our attempts to deploy an aqueous solution of Mg(Ac)₂ as a catalyst yield no detectable MAL (Fig. 4b). It suggests that Mg(Ac)₂ has no catalytic activity in this reaction independently, despite its Lewis acidity, highlighting the indispensable role of sec-amine center as the true catalytically active site for this transformative process. Additional characterizations of these as-synthesized catalysts were conducted using ¹⁵N and ¹³C NMR spectroscopy, with difference in ¹⁵N and ¹³C chemical shifts (Δδ) recorded against [HDEA]Pc IL as a reference. As depicted in Fig. 4e, the incremental rise in the molar ratio of Mg²⁺ to [HDEA]Pc IL leads to a proportional decrease of Δδ¹⁵N from 0 to −0.90 ppm, while the TOF exhibits a corresponding upward trend, spanning 0.67 to 1.36 h⁻¹. Whereas the Δδ¹³C climbs proportionally from 0 to 0.43 ppm and ln r_d exhibits a linear decline (Fig. 4f). These observations consolidate our conclusion that the Mg²⁺ chelation effect favors the electron-rich state of catalytically amine site, however, electron-deficient state of hydroxyl oxygen, thereby significantly improving catalytic activity and stability.

Fig. 4. Catalytic performance and mechanistic insights.

a, b Concentration profiles of (a) catalytically active amine species and (b) MAL catalyzed by Mg(Ac)₂, [HDEA]Pc IL, and MgIL-X (X = 0.25, 0.5, 0.75, 1.0) as a function of reaction time. c Effect of Mg²⁺ content in MgIL on the TOF, C_MAL, r_d and stability. d Comparison of catalytic performance among [HDEA]Pc IL and MgIL-X (X = 0.25, 0.5, 0.75, 1.0). e The relationship between the TOF and ¹⁵N chemical shift difference (Δδ¹⁵N = δ¹⁵N_MgIL − δ¹⁵N_{[HDEA]Pc IL}); ¹⁵N chemical shifts are provided in Fig. 2f. f The relationship between the ln r_d and ¹³C chemical shift difference (Δδ¹³C = δ¹³C_MgIL − δ¹³C_{[HDEA]Pc IL}); ¹³C chemical shifts are illustrated in Fig. 2e. g DFT calculations of the sec-amine site dissociation, formaldehyde (FA) activation, and MB⁺ enolization steps during the probe aldol condensation of formaldehyde with propionaldehyde catalyzed by [HDEA]Pc IL and MgIL. The embedded structures depict their respective transition states. The atoms in cyan, red, blue, yellow, and gray represent C, O, N, Mg, and H atom, respectively. h Distance of (H)O···C(N⁺) and corresponding energy for hydroxyl group rotation in IMI intermediate at various angles under the situation of hydrogen-bonding interaction (IMI-Pc⁻) and Mg²⁺ chelation effect (IMI-Mg). i The comparable relative energies required for rotation of hydroxyl group in IMI intermediate during intramolecular cyclization with hydrogen-bonding (IMI-Pc⁻) and chelation (IMI-Mg) interactions. The blue, green, and red segments in the pie chart respectively represent hydrogen-bonding interaction, van der Waals force, and repulsive interaction. The circular arrows indicate the rotation process. j Kinetic observation of generation of the deactivated byproduct 3-oxazolidineethanol.

DFT calculations subsequently unveil that the chelation of Mg²⁺ promotes the dissociation of catalytically active sec-amine center compared with pristine [HDEA]Pc IL, which is a crucial step for the following formaldehyde activation (Fig. 4g). It was also observed in other MILs containing various metal-ions that the enhanced dissociation of catalytically active sec-amine center is correlated to the increased Hirshfeld charge on the central nitrogen atom (Supplementary Fig. 23, and Supplementary Table 4). Further analysis of the Gibbs free energy barriers (ΔG^‡) of transition states involved in the critical steps of formaldehyde activation and MB⁺ enolization reveals that the determining barriers for [HDEA]Pc IL (16.0 kcal·mol⁻¹) and MgIL (15.3 kcal·mol⁻¹) are comparable (Fig. 4g, Supplementary Figs. 24–25, and Supplementary Tables 5–6). Thus, it can be concluded that the enhanced dissociation of sec-amine site predominantly contributes to the improved catalytic activity. Additionally, we also discovered that the hydroxyl group readily combines with the carbocation of IMI intermediate at a rotation angle of 150° to form undesired 3-oxazolidineethanol, primarily leading to catalyst deactivation²⁶, with an energy requirement of only 10.5 kcal·mol⁻¹ when the hydroxyl group is constrained by hydrogen bond with Pc⁻ anion (Fig. 4h, i, and Supplementary Table 7). By comparison, the relative energy dramatically rises to 68.1 kcal·mol⁻¹ when the hydroxyl groups are restrained by the Mg²⁺ chelation. So, we can conclude that the metal-ion chelation effect is significantly stronger than hydrogen-bonding interaction and thereby effectively hinders the undesired cyclization reaction, which is consistent with the in situ kinetic observations of 3-oxazolidineethanol generation (Fig. 4j) and explains the higher stability of MgIL.

Catalytic scale-up and extension

Given the excellent catalytic performance of MgIL-1.0 after controllable modulation, we first conducted the kilogram-scaled aldol condensation for probe MAL synthesis to demonstrate the application of this catalytic system in large-scale production. The scalable preparation of MgIL-1.0 can be easily achieved due to the convenient dissolution and chelation at ambient temperature and pressure in less than an hour, limited only by the size of reactor vessel (Fig. 5a). As illustrated in Fig. 5b and Supplementary Figs. 26–28, the MgIL-1.0 catalyst can be recycled for five runs consecutively without a loss of activity, which achieves an average conversion of 98.6% for propionaldehyde and 98.8% selectivity for MAL. Furthermore, we employed FT-IR and ¹H NMR to confirm the catalytically active amine component in catalytic system after multiple cycles of reaction, phase separation, evaporation, and recovery. Figure 5c demonstrates that the characteristic peaks of C = O and C–N bonds in the anion and cation of MgIL-1.0 stay consistent after recycling. The ¹³C NMR spectra indicate that the ¹³C signals of the catalytically active component are still the most prominent after five cycles, which is in accordance with those of fresh catalyst (Fig. 5d). Additionally, only a few insignificant peaks attributed to the oxazolidine species formed through intramolecular cyclization (catalyst deactivation) were observed, which are in agreement with the characteristic chemical shifts reported in literatures^59,60. Aside from the qualitative analysis, the quantitative results reveal that the contents of catalytically active amine component and Mg²⁺ in MgIL-1.0 remain virtually unchanged, with a high recovery ratio of around 97% (Fig. 5e).

Fig. 5. Reusability and applicability assessment of MgIL-1.0 catalyst.

a Photograph of catalyst preparation (embedded), kilogram-scaled synthesis, and phase separation systems. b Catalytic performance of MgIL-1.0 during recycling of the probe aldol condensation between the formaldehyde and propionaldehyde. c FT-IR spectra of the fresh and recycled MgIL-1.0 catalyst. d ¹³C NMR spectra of fresh and recycled MgIL-1.0 catalyst. R represents H–, CH₃–, and CH₂ = C(CH₃)– groups, corresponding to the oxazolidines derived from formaldehyde, propionaldehyde, and MAL, respectively. e Comparison of Mg and N contents in the theoretical, fresh, and recycled MgIL-1.0 catalyst. The error bars for the mean values represent the standard deviations calculated from three independent measurements. f Extensive applications of MgIL-1.0 in the synthesis of representative α, β-unsaturated aldehydes via aldol reaction. The products were identified through GC-MS analysis (Supplementary Figs. 29–36), while the stability was evaluated using ¹H NMR spectroscopy (Supplementary Figs. 37–47).

To further characterize the performance of this MgIL-1.0 catalyst in aldol condensation, we then investigated its applicability with substrates containing various electron-donating and withdrawing groups, such as alkyl, hydroxyl, furan, halogenated and benzoate groups (Fig. 5f). By increasing the alkyl chain length from 1 to 4 carbon atoms and introducing branching, the catalytic performance of MgIL-1.0 still shows a notable enhancement compared to pristine [HDEA]Pc IL, particularly for the catalytic stability improved by 1.4–3.1 times. It also demonstrates a significant improvement in overall performance on high-value fine chemicals applied as commercial fragrances^61,62, such as hydroxycitronellal and 3-(5-methyl-2-furyl)propionaldehyde. As for the aldehydes having electron-withdrawing groups like chlorine and phenyl substitution, the stability of MgIL-1.0 is improved by 1.2 to 1.7 times, with selectivity and TOF values increasing by 1.2 to 1.7 and 1.3 to 1.5 times, respectively. Notably, all of these above aldol reactions achieved over 90% selectivity for α, β-unsaturated aldehyde products, a level typically unattainable with conventional IL catalysts. The high catalytic selectivity achieved along with the enhanced TOF values and stability under MgIL-1.0 catalysis exclude the impacts of substituted group and steric hindrance, which suggests the excellent compatibility and overall performance of MgIL on the construction of C = C bonds via aldol reaction.

Discussion

In summary, we theoretically predicted and experimentally synthesized a series of MIL catalysts with tailored electronic characteristics for the efficient and stable synthesis of α, β-unsaturated aldehydes via aldol condensation at near-ambient condition. The electronic modulation mechanism is based on intramolecular inductive and coordination effects, along with the local environment of cation and anion resulting from metal-ion chelation. DFT calculations and in situ observations reveal the catalytic activity and stability of MIL are linearly correlated to the Hirshfeld charge of the central nitrogen (namely catalytically amine site) and hydroxyl oxygen atoms. While the TOF value of MIL shows a volcano-shaped trend with the increasing interaction between the metal-ion and [HDEA]Pc IL, which can be considered as a descriptor for MIL design. Comprehensive characterizations indicate that the electronic properties and local environment of the MgIL can be further controllably regulated by the content of Mg²⁺, which consequently reconfigures the nanoscale morphology and promotes the dissociation and stability of the catalytically active sec-amine center. As a result, the TOF value can be increased by 2.0-fold with catalytic deactivation rate suppressed by 89.6%, while the kilogram-scaled recycling pilot achieves 98.6% conversion and 98.8% selectivity for probe MAL synthesis. It also demonstrates excellent applicability and tolerance across various substrates with representative functional groups. In light of the impressive catalytic activity, stability, scale-up, and broad extension, this strategy represents a promising approach for sustainable transformation of low-carbon aldehydes into value-added chemicals by constructing new C = C bonds. Our findings also unveil the potential importance of electronic modulation during IL design and provide an avenue for the development of efficient and stable catalytic processes at near-ambient condition.

Methods

Chemicals

The chemical reagents and materials were purchased from commercial suppliers and utilized without additional purification. Diethanolamine (purity, ≥ 99%), propionic acid (purity, ≥ 99%), cesium acetate (purity, ≥ 99.9%), lithium acetate (purity, ≥ 99.99%), sodium acetate (purity, ≥ 99.99%), potassium acetate (purity, ≥ 99%), barium acetate (purity, ≥ 99.99%), magnesium acetate (purity, ≥ 98%), manganese acetate (purity, ≥ 98%), zinc acetate (purity, ≥ 99.99%), yttrium acetate hydrate (purity, ≥ 99.9%), formalin (concentration, 37 wt.% aqueous solution), propionaldehyde (purity, 97%), methacrolein (purity, ≥ 95%), butyraldehyde (purity, 99.5%), valeraldehyde (purity, ≥ 97%), iso-valeraldehyde (purity, 98%), hydroxycitronellal (purity, ≥ 98%), hydrocinnamaldehyde (purity, ≥ 95%), 3-(5-methyl-2-furyl)propionaldehyde (purity, ≥ 98%), ethanol (purity, ≥ 99.5%), 1,4-dioxane (purity, ≥ 99.0%), 4-hydroxy-2,2,6,6-tetramethyl-piperidinooxy (purity, ≥ 98%), 1, 2, 4, 5-tetrachlorobenzene (purity, ≥ 98.0%), deuterated dimethyl sulfoxide (DMSO-d₆, purity, d-99.9%) and deuterated chloroform (CDCl₃, purity, d-99.8%) were purchased from Aladdin Chemical Co., Ltd (Shanghai, China). Acetaldehyde (purity, 99.5%) and 5-chloropentanal (purity, 98%) were obtained from Macklin Biochemical Co., Ltd. (Shanghai, China). Deionized water was made in the laboratory.

Catalyst preparation

For the synthesis of [HDEA]Pc IL, equimolar amounts of DEA and HPc were mixed in a flask with magnetic stirring under ice bath. Typically, 42.5 g (0.4 mol) of DEA was combined with 127.7 g of water in a 500 mL three-neck flask. Under an argon atmosphere, 29.8 g (0.4 mol) of HPc was slowly added through a pressure-equalizing dropping funnel. The mixture was stirred for 1 h, resulting in a clear and colorless IL. While the MIL was synthesized by gradually adding the required amount of metal acetate to the [HDEA]Pc IL under ultrasonic agitation and vigorous stirring, followed by keeping at room temperature for 30 min. These obtained compounds are designated as MIL-X, where M represents the type of metal-ion, and X indicates the molar ratio of corresponding metal acetate to [HDEA]Pc IL. For the catalysts presented in Fig. 1c–e, the molar ratio of metal ions to [HDEA]Pc IL is 1:2. For the synthesis of MgIL-1.0 sample, 20 g (40 mmol) of the synthesized [HDEA]Pc IL was charged in a 100 mL three-neck flask. Then, 5.8 g (40 mmol) of magnesium acetate was gradually added under ultrasonic agitation and vigorous stirring. The reaction was continued for 30 min to yield a clear and transparent catalyst sample.

Catalyst characterizations

X-ray photoelectron spectroscopy (XPS) was performed on a ThermoFisher ESCALAB 250 Xi spectrometer utilizing Al-Kα radiation, while the analysis was conducted at a power of 200 W. All binding energies are referenced to the C1s peak at 284.8 eV for calibration.

¹H/¹³C/¹⁵N NMR spectra were obtained from a Bruker Avance ⅠⅠⅠ HD 600 MHz spectrometer at temperature of 25 °C, with CDCl₃ or DMSO-d₆ serving as the field frequency lock. The calibration of chemical shift was achieved by referencing the residual solvent peak or tetramethylsilane peak.

Isothermal titration calorimetry (ITC) measurements were carried out employing a Malvern MicroCal Auto-iTC200 calorimeter. A 20 mmol·L⁻¹ aqueous magnesium acetate solution was injected into a reaction cell containing 1 mmol·L⁻¹ [HDEA]Pc IL under the stirring at 800 r·min⁻¹ and 25 °C. The water titration without [HDEA]Pc IL served as the blank control. Titrations employed an initial delay of 60 seconds, followed by a 0.5 µL injection for equilibration but not for fitting. Subsequent 2 µL injections, each lasting 30 s, were administered at 120 s intervals for a total of 20 injections.

Fourier transform infrared (FT-IR) spectroscopy was carried out using a ThermoFisher Nicolet-380 spectrometer, the spectra were collected in the wavenumber ranging from 400 to 4000 cm⁻¹ with 32 scans at a resolution of 2 cm⁻¹. Samples were uniformly applied to potassium bromide tablets for measurements.

X-ray absorption near-edge structure (XANES) spectra of all samples were recorded at the Singapore Synchrotron Light Source. The powdered MgIL-1.0 sample was purified from the aqueous solution using the freeze-drying method. A pair of channel-cut Si (111) crystals was utilized in the monochromator, while the storage ring operated at an energy of 800 MeV, with an average electron current maintained below 200 mA.

Far infrared (FIR) spectroscopic characterizations were conducted on a Thermo Scientific Nicolet IS50 spectrometer equipped with a DLaTGS detector. The catalyst sample was first pretreated under ultrasonic for 10 min to ensure uniformity before introduction into the sample pool to create a liquid film with suitable thickness. Spectra were recorded in the range of 80–450 cm⁻¹ with a resolution of 4 cm⁻¹ and 128 scans.

Ionic conductivity tests were performed using a Mettler Toledo FE-38 conductivity meter with an LE-710 electrode. About 10 mL of catalyst was placed in a 15 mL jacketed kettle equipped with a thermostatic circulation pump. The electrode was positioned below the liquid surface, and conductivity was assessed once the IL attained a stable temperature. For each subsequent measurement, the temperature setting was adjusted accordingly. E_a is defined by the following formula of ln σ(T) = ln σ₀ − E_a/RT⁵⁵. σ(T) represents the conductivity at a specific temperature, σ₀ is pre-exponential factor, E_a is conductive activation barrier, R is gas constant of 8.314 J·mol⁻¹·K⁻¹, and T is absolute temperature in Kelvin.

Cryogenic transmission electron microscopy (Cryo-TEM) experiments were conducted using a thin film of MIL solution sample (2 µL) placed onto a supported grid at room temperature (Lacey Formvar/Carbon, 200 mesh, Cu; Ted Pella, Inc.). The preparation of these thin solution films is proceeded under controlled temperature and humidity conditions (97–99%) within a specially designed environmental chamber. After allowing the excess liquid to be blotted with filter paper for 2–3 s, the films were quickly vitrified by plunging them into liquid ethane cooled by liquid nitrogen at freezing point. The grid was then transferred to a Gatan 626 Cryo-holder via a Cryo-transfer device and subsequently to the Thermo Fisher FEI Talos F200C TEM instrument (200 kV). Direct imaging was performed at approximately −175 °C with a 200 kV accelerating voltage, utilizing images captured by a SC 1000 CCD camera (Gatan, Inc., USA). Data analysis was carried out using Digital Micrograph software.

Dynamic light scattering (DLS) analysis was conducted using a ZetaSizer Nano ZS90 equipped with a four-sided optical path sample cell. The sample was sonicated for 10 min before being transferred to the cleaned sample cell. Each sample was tested in triplicate.

The nitrogen content of both fresh and recycled MgIL-1.0 catalysts was determined using an Elementar Vario MACRO cube, while the magnesium content was quantified with an Agilent 5800 ICP-OES.

Experimental setup of in situ Raman spectrometer system

We developed an in situ Raman spectrometer system for real-time monitoring of chemical reactions. The system consists of a laser, an optical path enhancement system, an in situ reaction cell, and a spectrometer (Supplementary Fig. 4). A continuous-wave laser provides up to 0.5 W of power, effectively exciting the samples and generating reliable Raman scattering signals. To improve the Raman signal intensity, we implemented an integrating sphere system based on confocal focusing and right-angle beam enhancement principles, allowing multiple reflections of the pump light within the sample and significantly increasing the efficiency of Raman scattering signal collection. The in situ reaction cell features a jacketed structure for precise temperature control using a circulating temperature-regulating medium. With a volume of 3 mL and sapphire windows for laser transmission and Raman signal acquisition. Data acquisition utilized an Andor IVAC 316 LDC-DD CCD detector paired with an Andor SR500i spectrometer, equipped with a 1200-line grating for a resolution of 1.5 cm⁻¹. The Raman spectroscopic data were processed using Andor Solis software.

In situ Raman observation for aldol condensation

Aldol condensation catalyzed by IL and MIL was conducted on the in situ Raman spectrometer for real-time monitoring of the reaction. First, the Raman spectrometer was activated, and the CCD temperature was lowered to −60 °C. The kinetic acquisition mode was set to cover a wavelength range of 545–587 nm, with an exposure time of 0.2 s and 75 accumulations per acquisition and a total of 260 acquisitions. For example, to evaluate the catalytic performance of MgIL-1.0, 1.8 g of synthesized catalyst was accurately weighed and added to the in situ reaction cell, and the thermostatic circulation pump was activated to maintain a temperature of 0 °C inside the cell. Simultaneously, the pre-prepared aldehyde mixture was kept at a constant temperature in the circulation bath to minimize temperature fluctuations during the initial reaction phase. Following this, 0.8 g of the aldehyde mixture was measured using a syringe. The spectroscopy acquisition program was initiated, and the aldehyde mixture was rapidly injected into the in situ reaction cell, followed by multiple aspirations to ensure thorough mixing of the system.

The concentration of MAL and catalytically active amine species was determined through the quantitative analysis of the characteristic peak at 1693 cm⁻¹ (C = O bond) and 1102 cm⁻¹ (C–N bond), respectively. TOF value was calculated by the following formula: TOF = r_Pro./C_Cat., where r_Pro. represents the formation rate obtained from linear fitting of product concentration at the initial stage of reaction, C_Cat. represents the initial concentration of catalyst. C_MAL is defined by the MAL concentration measured at the reaction time of 60 min. Stability is defined by the formula of Stability (%) = 1 − Q, where Q indicates the catalyst consumption percentage after 60 min of reaction. r_d was calculated from the linear fitting of catalytically active amine species concentrations at the initial reaction stage.

Catalytic scale-up and recycling

A total of 540.0 g of MgIL-1.0 was introduced into a 1 L five-necked jacketed reactor using a feed pump, which was equipped with a condenser and an electronic thermometer inserted below the liquid level. Mechanical stirring was initiated at 400 rpm, and a thermostatic circulation bath was activated to maintain the temperature at 40 °C. The feed solution was prepared by thoroughly mixing 203.7 g of formaldehyde, 150.3 g of propionaldehyde, and 0.35 g of 4-hydroxy-2,2,6,6-tetramethyl-piperidinooxy. This mixture was then fed into the reactor by the pump at a drop rate of 17.7 g·min⁻¹ after calibration with the feed solution (Supplementary Fig. 26). After reaction for 10 min, the mixture was transferred to a separatory funnel with a cooling jacket using another pump and then cooled to 10 °C. Once the liquid level stabilized, the aqueous and organic phases were separated by slowly turning on the valve (Supplementary Fig. 27). Both phases were analyzed by gas chromatography using 1, 4-dioxane as an internal standard (Supplementary Fig. 28). The organic phase was subsequently subjected to rotary evaporation to remove excess water before being recycled for the next run.

Catalytic extension assessments

Catalytic extension was performed in a 50 mL three-necked flask equipped with a stirrer. For the evaluation of catalytic selectivity and TOF, the catalyst was added into the reactant mixtures and the reaction was conducted at 25 °C for 3–5 min with a formaldehyde/substrate/catalyst molar ratio of 8:4:1. For the substrates of butyraldehyde, valeraldehyde, and iso-valeraldehyde, the molar ratio was maintained at 4:4:1. Then the mixture was diluted with specified amounts of ethanol and 1, 4-dioxane. The products were confirmed using gas chromatography–mass spectrometry (GC–MS) (QP2020, Shimadzu). Quantitative analysis was performed on an Agilent 7890 A GC equipped with a flame ionization detector and a DB-624 column (60 m × 0.32 mm × 1.8 μm). The relative mass correction factor was determined using a standard mixture, while the effective carbon number method was used for the products without a standard sample. To assess catalytic stability, the reactants and catalyst were mixed in an equimolar ratio of formaldehyde, substrate, and catalyst, and maintained at 25 °C for 1 h to achieve equilibrium. Upon completion, the reaction mixture was diluted with ethanol and then dissolved in DMSO-d₆ for ¹H NMR analysis.

DFT calculations

All calculations in this study were carried out using the Gaussian 16 software package⁶³. The structural optimization was carried out using B3LYP hybrid density functional with 6-311 + G (d, p) basis set for C, H, O, N, Na, Li, and Mg atoms, and SDD pseudopotential basis set for K, Mn, Y, Cs, and Ba^64,65. The dispersion correction was also considered. Vibrational analysis was performed for the optimized structures, and no imaginary frequency was found. In order to obtain more accurate electronic energy, the M06-2X functional, which is suitable for calculation of non-covalent interactions and reasonable description of dispersion effects, was employed along with the def2-TZVP basis set in the single-point energy calculation^66,67. In the calculations involving the reaction, water was set as the implicit solvation model. The Integral Equation Formalism of Polarizable Continuum Model was used for structural optimization and vibrational analysis, while the Solvation Model Based on Density was used for single-point energy⁶⁸. The transition state (TS) was confirmed to have only one imaginary frequency, and the intrinsic reaction coordination analysis was used to obtain the initial state (IS) and final state (FS) structures of the connected TS. The Multifwn software was applied to calculation of electronic properties based on wavefunctions⁶⁹. The VMD program was used for graphic visualization⁷⁰. The Gibbs free energy was calculated by Shermo software⁷¹. ΔG^‡ was calculated by the formula of ΔG^‡ =  G(TS)  −  G(IS). Here, G(TS) and G(IS) are the calculated Gibbs free energies of TS and IS in each elementary step, respectively. The temperature and pressure were set as 313.15 K and 1 atm, respectively. The interaction energy (ΔE_M–IL) between the metal acetates and [HDEA]Pc IL was calculated using the formula of ΔE_M–IL = E_MIL − E_M − E_IL. Here, E_MIL refers to the energy of MIL, E_M represents the energy of metal acetate, and E_IL indicates the energy of [HDEA]Pc IL. In the presence of metal acetate, the interaction energy between HPc and DEA (ΔE_{M(DEA∙∙∙HPc)}) is defined by ΔE_{M(DEA∙∙∙HPc)} = E_MIL − E_DEA–M − E_HPc–M + E_M⁷². Here, E_MIL represents the energy of MIL, E_DEA–M refers to the energy of the combined DEA and metal acetate, E_HPc–M denotes the energy of the combined HPc and metal acetates, and E_M is the energy of metal acetate.

Author contributions

T.H. Zhang designed the major experiments and analyzed the data. Y. Tian conducted the DFT calculations for the mechanistic investigation. C. Zhang synthesized the catalysts and evaluated their catalytic performance. T.T. Yan analyzed the XANES spectra. H.W. Yan, G.L. Zhang, J.Li and Z.X. Li provided insights into the characterization results. G. Wang, C.S. Li and S.J. Zhang contributed to the project administration, formal analysis and research guidance. The manuscript was written by T.H. Zhang, Y. Tian and G. Wang with input from all authors.

Peer review

Peer review information

Nature Communications thanks Sergio Rossi and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The source data generated in this study have been deposited in the Figshare database under accession code 10.6084/m9.figshare.29502947.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-63630-9.