Introduction

Sustainable and effective methods for the utilization of various alternative clean energy resources are crucial and become a common challenge for all mankind. Ferroelectrics are a type of important multifunctional materials with inherent spontaneous polarization that can be reoriented under an external electric field1,2. Besides the ferroelectricity, ferroelectrics also simultaneously have excellent piezoelectric, dielectric, and pyroelectric properties, making them widely used in many technical devices, such as non-volatile memories, sensors, actuators, capacitors, transducers, and detectors3,4,5,6,7,8,9. On the other hand, the inherent spontaneous polarization in ferroelectrics endows them with built-in electric fields, which could effectively separate external free electrons and holes, greatly facilitating the participation of surface redox reactions10,11,12,13,14,15. Furthermore, ferroelectrics with the same spontaneous polarization vector in nano size, that is, the ferroelectric domain, could more efficiently absorb external charges and release them with the change of polarization under external stimuli like stress to further activate catalytic reactions16,17,18,19,20. In recent years, ferroelectrics have attracted ascending attraction as catalysts, showing multiple applications in water splitting, pollutant degradation, and other catalytic organic synthesis reactions21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37. Among them, traditional inorganic ferroelectrics (BaTiO3, BiFeO3, NaNbO3, Bi4Ti3O12, Bi2WO6, etc) with good ferroelectric-related performances have been widely studied for their catalytic potential38,39,40,41,42,43. For example, Ito’s group pioneeringly reported the ferroelectric BaTiO3 to catalyze organic reductive coupling reactions under mechanical ball milling conditions, which ensured the scaling up for industrial manufacture44. However, the preparation of traditional inorganic ferroelectric catalysts generally needs a high-temperature and complex process. On this account, molecular ferroelectrics have been attracting enormous interest in the last decades due to their simple low-temperature solution processing as well as their excellent ferroelectric-related performances such as large spontaneous polarization, high curie temperatures, high piezoelectric coefficients, and multipolar axis, represented by the organic-inorganic perovskite molecular ferroelectric TMCM–CdCl345,46,47,48,49,50,51,52. Moreover, molecular ferroelectric, as a valuable complement to inorganic counterparts, reveals the great potential for catalysis profiting from their diversified species, property adjustability, and reaction adaptation, which are more likely to facilitate catalyzing specific organic reactions. However, molecular ferroelectrics remain blank in terms of catalysis, and chemical reactions mediated by ferroelectric materials are still in their infancy.

Quinolines are prevalent in biologically active natural products and represent privileged moieties in medicinal chemistry. Their pharmacological importance and utility as the cornerstone of organic synthesis have encouraged extensive research activities to synthesize appropriately substituted quinoline rings53,54,55,56. Numerous synthetic protocols to elaborate this molecular scaffold have been developed over the years, including the name reactions of Skraup, Conrad–Limpach, Friedländer, and the recently developed transition-metal-catalyzed C–H activation, etc.57,58,59,60,61,62,63,64,65,66,67,68. Yi et al. developed an organic-inorganic hybrid cobalt(III) complex-catalyzed [3 + 2 + 1] cyclization of anilines and terminal alkynes by using dimethyl sulfoxide (DMSO) as the synthon for building the quinoline backbone, which led to the specific synthesis of privileged substituted quinolines in one-pot and shows a significant amount of attention69,70,71. Although continuously alternative approaches are emerging, some of them suffer from significant limitations in terms of expensive and handful of catalysts, harsh conditions, or poor regioselectivity. Accordingly, simple and efficient methodologies for the direct construction of the quinoline skeleton are still highly desirable and present an urgent need.

Herein, we report the introduction of molecular ferroelectric crystal TMCM–CdCl3 for catalytic synthesis of the substituted 3-arylquinolines in one shot via [3 + 2 + 1] annulation of anilines and terminal alkynes by using DMF as both the solvent and the carbon source. TMCM–CdCl3 crystal possesses multiaxial ferroelectricity with 12 equivalent polarization directions and superior piezoelectricity with a large d33 value of 383 pC/N and a decent electromechanical coupling factor k33 of 0.48372. Under ultrasonicated conditions, TMCM–CdCl3 demonstrates evident catalytic activity in contrast to the controls of inorganic ferroelectric BaTiO3, piezoelectric ZnO, and centrosymmetric structural analogs TM–CdCl3 (TM = tetramethylammonium) and TMFM–CdCl3 (TMFM = trimethylfluoromethylammonium). The proposed ferroelectric catalytic mechanism is mainly attributed to the free charges released from the surface of ferroelectric domains in nano size that quickly participate in the vital redox reaction (Fig. 1). To the best of our knowledge, this finding demonstrates the application of molecular ferroelectrics in catalysis, and it could have a significant implication both in the synthetic and catalysis fields.

Fig. 1: Ferroelectric catalytic annulation for 3-arylquinolines by molecular ferroelectric crystal TMCM–CdCl3 under ultrasonic condition.
figure 1

a The periodic polarization change of the molecular ferroelectric crystal under ultrasonic conditions produces the free charges on the surface. b Annulation of arylamine, arylacetylene, and DMF was catalyzed by molecular ferroelectric TMCM–CdCl3 for 3-arylquinolines. The atomic labeling denotes that the C3 and C4 atoms of 3-arylquinolines belong to arylacetylene, the N1 originates from arylamine, and DMF serves as a C2 source. c The possible mechanism shows the redox reactions that occurred on the surface of ferroelectric domains, in which different colors represent ferroelectric domains with different orientations and the free charges released with the periodic change of polarization.

Full size image

Results and discussion

Crystal structures and phase transition

The synthesis of ferroelectric crystal TMCM–CdCl3 is described in the section of Methods. In the preliminary designed experiment, the annulation reaction of aniline and phenylacetylene catalyzed by TMCM–CdCl3 using DMF as solvent was performed for the synthesis of indoles under sonication. However, the unexpected 3-phenylquinoline was obtained and unambiguously confirmed by the single-crystal structure (Supplementary Fig. 1). This implies that the C3 and C4 atoms of 3-phenylquinoline belong to phenylacetylene, the N1 originates from aniline, and DMF serves as a C2 source for the construction of quinoline rings. To confirm the unique ferroelectric contribution, a series of control experiments were carried out (Supplementary Table 1). The organic ligand TMCM chloride47, CdCl2, traditional ferroelectric BaTiO347, piezoelectric ZnO47, and non-ferroelectric structural analogs such as TM–CdCl346 and TMFM–CdCl346 have shown that none of these materials could catalyze annulation for 3-arylquinolines. When structural analogs of molecular ferroelectric solid solution [TMFM-TMCM]CdCl3 were introduced46, they were positive to the cyclization but demonstrated low catalytic activity. As a result, molecular ferroelectrics could be the potential catalyst for the annulation of quinolines, and this strategy is worth exploring in depth.

To exclude the negative effect of the reaction solvent on the ferroelectricity, we first recrystallized the sample in DMF by slowly evaporating the solvent at room temperature. The TMCM–CdCl3 crystal recrystallized from DMF at 300 K belongs to a polar Cc space group (Laue point group m), corresponding to the ferroelectric phase, where the TMCM cations are in a fully static-ordered state (Fig. 2a and Supplementary Table 2). At 423 K, TMCM–CdCl3 transfers to a centrosymmetric P63/mmc space group (Laue point group 6/mmm) with 12-fold orientational disordered TMCM cations in structure, corresponding to the paraelectric phase (Fig. 2b and Supplementary Table 2). Therefore, TMCM–CdCl3 recrystallized from DMF undergoes a 6/mmmFm type ferroelectric phase transition. The differential scanning calorimetry (DSC) shows that the phase transition temperature of the recrystallized TMCM–CdCl3 is at 400 K (Fig. 2c). The significant symmetry breaking in second-harmonic generation (SHG) response and dielectric constant with typical λ-type anomaly further demonstrate that the TMCM–CdCl3 recrystallized from DMF possesses the same intrinsic reversible ferroelectric phase transition to the protogenetic crystals46,47 (Fig. 2d and Supplementary Fig. 2). Furthermore, the thermogravimetric analysis (TGA) shows the thermal stability of recrystallized TMCM–CdCl3 up to about 580 K (Supplementary Fig. 3).

Fig. 2: Crystal structures and phase transition of TMCM–CdCl3 recrystallized from DMF.
figure 2

Packing structures at 300 K (a) and 423 K (b). The green (chloride atoms) and blue (nitrogen atoms) clouds indicate the disordered states of organic molecules in the paraelectric phase. H atoms were omitted for clarity. c DSC curve. d Temperature-dependent SHG intensity and dielectric constant in a heating run.

Full size image

Ferroelectricity

Besides, piezoresponse force microscopy (PFM) characterizations of the thin film of TMCM–CdCl3 recrystallized from DMF were also performed. Figure 3a–c shows out-of-plane (OP) PFM images as well as simultaneously captured topographic images. Clear ferroelectric domains can be observed. Dark lines in the amplitude image indicate the location of the domain walls (Fig. 3a), and different color tones in the phase image indicate different polarization directions, showing the multiaxial ferroelectric nature (Fig. 3b). There is no correlation between the domain structure and the surface morphology (Fig. 3c). Next, tip voltages of −50 V and +50 V were applied to a region of a single-domain state successively, which led to the generation of a new domain (Fig. 3d, e) and the new domain switched back (Fig. 3f, g). Figure 3h shows the topography of the same region without any damage after domain switching measurements. Further, Fig. 3i shows the local PFM switching spectroscopy. The square phase-voltage hysteretic loops (up) and butterfly-shaped amplitude-voltage loops (down) evidence the ferroelectric switching. We also performed domain switching measurements on the pressed-powder pellet sample of TMCM–CdCl3 grown from DMF (Supplementary Fig. 4). The topography image shows a surface morphology of polycrystalline particles (Supplementary Fig. 4a). It can be seen that in the initial state, some small domains are randomly distributed in this region (Supplementary Fig. 4b, c). Then, a concentric box-in-box domain pattern was produced by scanning the center region, first, with a tip under a voltage of +60 V (Supplementary Fig. 4e, f, large purple square) and, then, under a negative tip voltage of –60 V (Supplementary Fig. 4h, i, small purple square). During the poling processes, the surface topography shows no damage (Supplementary Fig. 4d, g). These results indicate that the polarization in TMCM–CdCl3 thin film and powder sample can both be reversed under an external electric field, confirming the ferroelectric nature of TMCM–CdCl3. At the same time, we compare the piezoresponse of TMCM–CdCl3 thin film grown in DMF with that grown in H2O. We drive each film across resonance frequency using PFM, where the resonance peak is comparable between two thin films (Supplementary Fig. 5a). The resonance curves were fitted very well by the simple harmonic oscillator (SHO) model73, with which the intrinsic piezoresponse can be derived by correcting the resonance amplification using the quality factor. Such resonance measurements have been repeated under different voltages and different positions. Then the intrinsic piezoresponse was plotted against the driving voltage for two films, which shows good linearity (Supplementary Fig. 5b). The slopes of the two films are almost equal, indicating the well-maintained piezoelectric response. Therefore, TMCM–CdCl3 crystallizing in DMF could maintain good ferroelectricity and piezoelectricity, which provide significant evidence for the ferroelectric catalytic annulation for quinolines.

Fig. 3: Ferroelectricity of TMCM–CdCl3 recrystallized from DMF.
figure 3

a–c As-grown domain structures observed in TMCM–CdCl3 thin film: OP PFM amplitude (a), phase (b), and corresponding topography (c) images. d–g OP PFM amplitude d, f and phase (e, g) images after applying a tip voltage of −50 V to a single-domain region (d, e), and after applying an opposite tip voltage of +50 V (f, g). h Topography image of the same region after domain switching operation. The regions shown in the purple dashed boxes are the written domains obtained after applying the tip voltages. All the panels have the same scale bar of 1 μm. i Local PFM switching spectroscopy.

Full size image

Ferroelectric catalysis

To explore the optimal ferroelectric catalytic conditions, we conducted a series of control experiments using aniline, phenylacetylene, and DMF as templates shown in Supplementary Tables 1 and 3–4. In this reaction, tetrabutylammonium acetate (nBu4N(OAc)) acts as an electrolyte to facilitate the transfer of charges generated by molecular ferroelectrics under ultrasonic conditions. Without the engagement of electrolytes, the yield decreased significantly. When a mixture of DMF and water was used as the solvent, an obvious decrease in yield was obtained (entries 2–6 in Supplementary Table 3), revealing that the presence of water was not conducive to the reaction and that it may involve dehydration. It has been reported that DMSO can also be used as a carbon source for the synthesis of quinolines57,58,59,60,61,62,63,64,65. However, entry 7 (Supplementary Table 3) shows that DMSO is unsuitable in this case, indicating that the ferroelectric catalytic mechanism should be different and the carbon source might be from the aldehyde group. Temperature is also a significant factor. We observed that the interior temperature of the reaction was 90 °C upon sonication for 9 h. It should be noted that the ferroelectric catalyst TMCM–CdCl3 is still in the ferroelectric phase at this temperature. Other conditions being equal, when the reaction was performed in a water bath to maintain the temperature at room temperature under the sonication, only trace amounts of the product were detected (Supplementary Table 4). In comparison, the experiments conducted only with stirring without ultrasonic assistance show traces even when heated at 120 °C. As a result, the optimal reaction condition was found to be 1 mmol of aniline, 1 mmol of phenylacetylene, and 0.2 mmol of tetrabutylammonium acetate in 10 mL of DMF under sonication in the presence of excess amount of ferroelectric catalyst TMCM–CdCl3 (entry 1 in Supplementary Table 1). After separating the crude product, TMCM–CdCl3 was recycled and reused under identical conditions. Supplementary Fig. 6 shows the powder X-ray diffraction (PXRD) pattern of TMCM–CdCl3 that was repeatedly used for catalytic reaction with no obvious change, indicating the high stability of TMCM–CdCl3 while maintaining consistent catalytic activity. With the optimized reaction conditions in hand, we explored the scope of substrates for 3-arylquinolines and the ferroelectric catalytic applicability of TMCM–CdCl3. In Fig. 4, several kinds of substituted aniline and phenylacetylene were used to give targets with high regioselectivity (3b–3r). The corresponding nuclear magnetic resonance (NMR) and mass spectra are in the Supplementary Information (Supplementary Figs. 7–42). To view the unambiguous structures more directly, we tried to obtain the crystal structures of 3-arylquinolines. Partly obtained 3-arylquinolines were acidified into their chloride salts, which retain their structural attributes. As shown in Supplementary Figs. 43, the crystal structures of the 3a–3b, 3d, 3g, and 3l in their salts and the 3p–3r with their intrinsic structural model were demonstrated, and the details of crystal parameters were listed in Supplementary Tables 5 and 6.

Fig. 4: Ferroelectric catalysis.
figure 4

Molecular ferroelectric catalyzed synthesis of 3-arylquinoline derivatives. The yields are of the isolated products.

Full size image

To clarify the reaction pathway, control reactions were carried out as shown in Supplementary Fig. 44. We chose DEF (N,N-diethylformamide) and DMA (N,N-dimethylacetamide) as an alternative to DMF, 3a could be only isolated from the one with DEF, denoting that C2 is indeed derived from the aldehyde group. When radical scavenger (2,2,6,6-tetramethylpiperidin-1-yl)oxyl (TEMPO, 3 equiv.) was added, only trace product was detected, indicating there are free radicals present and involved in this reaction. Then 2-styrylanilinewas synthesized as a substrate to react with DMF under standard conditions. The result showed that 3a could be detected by gas chromatograph-mass spectrometer (GC–MS), revealing that 2-styrylaniline is a possible intermediate. Besides, the multiaxial nature makes the domain structures of TMCM–CdCl3 susceptible to external force. We examined the changes in domain structures of TMCM–CdCl3 crystal before and after ultrasonic treatment. The samples used are cut sheet crystals. Before the ultrasonic treatment, most of the smooth surface area of the TMCM–CdCl3 crystal was in a single-domain state, except for a few sparse stripe domains (Supplementary Fig. 45a, up row). In addition, some small needle-like domains can be observed on the side of the crystal (Supplementary Fig. 45a, bottom row). After ultrasonic treatment, the crystal breaks into fragments, and a large number of needle-like domain structures can be observed by random detection of the surface of the fragments (Supplementary Fig. 45b). These results indicated that the domain structures of TMCM–CdCl3 can be changed by ultrasonic treatment, which contributes to the release of free charges from the surface of ferroelectric domains. Based on the above experiments, a plausible mechanism is outlined in Supplementary Fig.46. Specifically, aniline is oxidized with the concomitant loss of a proton to form N-radical species I, which then resonates to liberate the C-radical species II. Next, the radical addition between II and phenylacetylene is proceeded to form C-radical species III. Subsequently, intermediate III reacts with DMF to form intermediate V, which then experiences intermolecular cyclization accompanied by dehydration and forms intermediate VI. The dehydration could be supported by the control studies of entries 2–6 (Supplementary Table 3), in which the addition of water in the reaction shows an obvious inhibition. Finally, VI is reduced with the acceptance of electrons released from the surface of the molecular ferroelectric catalyst TMCM–CdCl3, and dimethylamine is ejected to form the target quinoline 3a. Besides, we introduced deuterated DMF-d7 to this reaction to further support the reaction mechanism. The final isolated product is 3a, not a deuterated one in the C2 position, and the liquid chromatograph-mass spectrometer (LC–MS) detection also revealed the residue of deuterated dimethylamine (Supplementary Figs. 47 and 48). Therefore, the substantial contribution of TMCM–CdCl3 lies in the effective release of free charges from the surface of ferroelectric domains subjected to the periodic ultrasonic treatments, which then participates in the electron loss, oxidation, generation of free radicals, and reduction process in the cyclization reaction, further confirming the important role of molecular ferroelectric catalyst in the entire catalytic cyclization process.

In summary, we have presented the introduction of the molecular ferroelectric crystal of TMCM–CdCl3 as a catalyst to the direct construction of 3-arylquinolines under ultrasonicated conditions via one-pot [3 + 2 + 1] annulation of anilines and terminal alkynes using solvent DMF as the carbon source. Significantly, well ferroelectricity and piezoelectricity of TMCM–CdCl3 could be maintained after the recrystallization in DMF. Compared to centrosymmetric analogs and inorganic piezoelectrics, TMCM–CdCl3 shows apparent catalytic activity. Through the mechanistic investigation, a plausible pathway was proposed, in which the free charges released from the surface of the ferroelectric domains in nano size with the periodic change of polarization quickly interact with the substrates in the solution to facilitate the decisive redox catalysis reactions. Based on the groundbreaking features of molecular ferroelectric catalysis with simple, cheap and valuable catalytic effects, this work would significantly promote the applications in the field of molecular ferroelectrics and also throw light on the advances in catalysis science and synthetic utility. Although the yield of 3-arylquinolines by molecular ferroelectric catalysis TMCM–CdCl3 is not high currently, we will continue to investigate and optimize factors based on this work. We expect to publish the progress in due course.

Methods

Materials

Aniline, p-anisidine, 4-isopropylaniline, phenylacetylene, 4-ethynyltoluene, 4-tert-butylphenylacetylene, 2-ethynyl-naphthalene, tetrabutylammonium acetate, N,N-dimethylacetamide (DMA), N,N-diethylformamide (DEF) were purchased from Macklin company. Diethyl ether, DMF, petroleum ether (PE), and ethyl acetate (EA) were purchased from Energy Chemical, Adamas company, and used as the solvent. We prepared the molecular ferroelectric crystal TMCM–CdCl3 according to the literature procedures46,47. Specifically, TMCM chloride (10 mmol, 1.43 g) and CdCl2 (10 mmol, 1.83 g) were mixed in aqueous solution (20 ml). Transparent single crystals of TMCM–CdCl3 were collected by slow evaporation of the solution at room temperature (yield 81%).

Synthesis procedure

In a 15 mL sealed tube, 1 mmol of aniline, 0.2 mmol of tetrabutylammonium acetate, and 1 mmol of phenylacetylene were dissolved in 10 mL of DMF, followed by the addition of 3 g of molecular ferroelectric TMCM–CdCl3 powders as a catalyst, in which 1.4 g of TMCM–CdCl3 was undissolved at room temperature. The reaction was carried out on FS-250N from ShengXi Ultrasonic Instrument (Shanghai, China). The ultrasonic probe was placed into the solution with an ultrasonic frequency of 20 kHz and a power of 67 W for 9 hours at a pulse mode operation for the 20s–on–10s-off cycle (Supplementary Fig. 49). The progress of the reaction was monitored by thin layer chromatography (TLC). After the reaction was terminated, the reaction mixture was then cooled to room temperature, and the unreacted solid catalyst was filtered off. The filtrate was washed with water, extracted with ethyl acetate, and dried. The residue was dispersed in 100 mL water and extracted with ethyl acetate (100 mL × 3). The organic layer was collected, dried over with sodium sulfate, filtered, and concentrated. The crude product was subsequently purified by column chromatography on silica gel (200–300 meshes) using a mixture of PE/EA (20:1) as the eluent. The crystal structures of 3a–3b, 3d, 3g, 3l in their salts and 3p–3r were presented in the Supplementary Information.

NMR spectra and gas chromatograph-mass spectrometer (GC–MS) analysis

NMR spectra were recorded on Bruker Switzerland AG instrument and calibrated by using residual undeuterated chloroform (δ, 1H NMR = 7.26). GC–MS analysis was performed on Shimadzu GCMS-QP2021 SE. The ultrasonicator was carried out on FS-250N from ShengXi Ultrasonic Instrument (Shanghai, China).

Single-crystal X-ray crystallography and powder X-ray diffraction (PXRD)

Single-crystal X-ray diffraction data were measured using a Rigaku XtalAB Synergy diffractometer (Cu-Kα radiation λ = 1.54148 Å). Data collection, cell refinement, and data reduction were performed using Rigaku CrystalClear 1.3.5. The data collection and structure refinement of these crystals are summarized in Supplementary Tables 5 and 6. PXRD data were measured using a Rigaku D/MAX 2000 PC X-ray diffraction system with Cu-Kα radiation in the 2θ range of 5°–40° with a step size of 0.02°.

DSC, dielectric, TGA, and SHG measurements

DSC measurements were recorded on a PerkinElmer DSC 6000 instrument with a heating rate of 20 K/min under a nitrogen atmosphere. Dielectric permittivity was measured with a TH2828A impedance analyzer. The conductive silver adhesive applied to the surfaces of samples was used as top and bottom electrodes. The thermogravimetric analysis (TGA) curves were measured by PerkinElmer TGA8000 under a nitrogen atmosphere. SHG experiments were performed on an unexpanded laser beam with low divergence (pulsed Nd:YAG at a wavelength of 1064 nm, 5 ns pulse duration, 1.6 MW peak power, 10 Hz repetition rate). The instrument model is Ins1210058, INSTEC Instruments, while the laser is Vibrant 355 II, OPOTEK.

PFM characterization and thin film preparation

The surface morphology, domain imaging and domain switching were studied by using commercial piezoelectric force microscopy (PFM, Oxford Instruments, MFP-3D) in single frequency mode. Commercial silicon tips coated with conductive Pt/Ir (PPP-EFM-W, NANOSENSORS) were used for PFM at typical contact resonance frequencies of 300–350 kHz for out-of-plane PFM and a typical drive amplitude of VAC = 5 V. As for the preparation of thin films of TMCM–CdCl3 recrystallized from DMF, the precursor solution was prepared by dissolving 50 mg of the TMCM–CdCl3 crystals in 1 mL of DMF. Then, 20 μL of precursor solution was spread on a clean indium-doped tin oxide (ITO) glass substrate (1 × 1 cm2). The solution was then slowly evaporated at 298 K to obtain good thin-film crystals for the PFM characterization.