Combination of Knoevenagel Polycondensation and Water-Assisted Dynamic Michael-Addition-Elimination for the Synthesis of Vinylene-Linked 2D Covalent Organic Frameworks.

Vinylene-linked two-dimensional conjugated covalent organic frameworks (V-2D-COFs), belonging to the class of two-dimensional conjugated polymers, have attracted increasing attention due to their extended π-conjugation over the 2D backbones associated with high chemical stability. The Knoevenagel polycondensation has been demonstrated as a robust synthetic method to provide cyano (CN)-substituted V-2D-COFs with unique optoelectronic, magnetic, and redox properties. Despite the successful synthesis, it remains elusive for the relevant polymerization mechanism, which leads to relatively low crystallinity and poor reproducibility. In this work, we demonstrate the novel synthesis of CN-substituted V-2D-COFs via the combination of Knoevenagel polycondensation and water-assisted dynamic Michael-addition-elimination, abbreviated as KMAE polymerization. The existence of C=C bond exchange between two diphenylacrylonitriles (M1 and M6) is firstly confirmed via in situ high-temperature NMR spectroscopy study of model reactions. Notably, the intermediate M4 synthesized via Michael-addition can proceed the Michael-elimination quantitatively, leading to an efficient C=C bond exchange, unambiguously confirming the dynamic nature of Michael-addition-elimination. Furthermore, the addition of water can significantly promote the reaction rate of Michael-addition-elimination for highly efficient C=C bond exchange within 5 mins. As a result, the KMAE polymerization provides a highly efficient strategy for the synthesis of CN-substituted V-2D-COFs with high crystallinity, as demonstrated by four examples of V-2D-COF-TFPB-PDAN, V-2D-COF-TFPT-PDAN, V-2D-COF-TFPB-BDAN, and V-2D-COF-HATN-BDAN, based on the simulated and experimental powder X-ray diffraction (PXRD) patterns as well as N2 -adsorption-desorption measurements. Moreover, high-resolution transmission electron microscopy (HR-TEM) analysis shows crystalline domain sizes ranging from 20 to 100 nm for the newly synthesized V-2D-COFs.

Introduction

Two‐dimensional π‐conjugated covalent organic frameworks (2D c‐COFs), which belong to the class of crystalline 2D conjugated polymers with in‐plane π‐conjugation and regular 2D framework structures, are attractive materials for potential applications in optoelectronics, spintronics, photocatalysis, and energy storage.[ , , , , , , , , , , ] For the solution synthesis of 2D c‐COFs, dynamic covalent chemistry, which can form‐break‐reform covalent bonds under a dynamic equilibrium (error‐correction process), is indispensable for forming thermodynamically controlled crystalline frameworks.[ , ] So far, the most reported 2D c‐COFs are limited to the imine‐linked[ , , , , ] and triazine‐linked 2D c‐COFs.[ , , ] However, due to the high polarization of C=N bond, imine‐ and triazine‐linked 2D c‐COFs exhibit a moderate electron delocalization with poor conjugation.[ , , , , ] Moreover, the imine‐linked 2D c‐COFs generally suffer from poor chemical stability, which hinders their potential for robust applications.[ , ]

Vinylene‐linked 2D covalent organic frameworks (V‐2D‐COFs, or vinylene‐linked 2D conjugated polymers) have recently attracted arising attention because of the robust and superb π‐conjugated framework structures with high chemical stability.[ , , , , , ] Several synthetic methodologies, including Knoevenagel‐,[ , ] other Aldol‐type‐,[ , , , ] and Horner–Wadsworth–Emmons based 2D polycondensation approaches, have been developed. Particularly, several CN‐substituted V‐2D‐COFs have been successfully synthesized via Knoevenagel polycondensation, and demonstrated superior performance as magnetic materials, organic cathodes for lithium‐ion batteries, luminescence,[ , , ] and photocatalysis,[ , , , ] compared with their corresponding imine‐linked 2D COF analogs. Nevertheless, the mechanism understanding for the formation of crystalline CN‐substituted V‐2D‐COFs remains elusive. It is still challenging to control the reversibility of the Knoevenagel 2D polycondensation, which is essential to enhance the crystallinity and reproducibility for the synthesis of V‐2D‐COFs.[ , ] Therefore, it is highly desirable to achieve a deep insight into the polymerization mechanism, including the driving force for the error‐correction process and controlled reaction conditions for synthesizing highly crystalline CN‐substituted V‐2D‐COFs.

In this work, we report a novel and highly reproducible strategy—the combination of Knoevenagel polymerization and water‐assisted dynamic Michael‐addition‐elimination (classified as KMAE polymerization)—for the synthesis of highly crystalline CN‐substituted V‐2D‐COFs (Figure 1a). A dynamic C=C bond exchange of model reaction of diphenyl‐acrylonitrile (M1) and (4‐methylphenyl)‐acetonitrile (M3) toward (Z)‐3‐phenyl‐2‐(p‐tolyl)acrylonitrile (M6) and benzonitrile (M2) is firstly demonstrated via in situ high‐temperature (HT) NMR investigations (Figure 1b). In particular, the intermediate 2,3,4‐triphenylpentanedinitrile (M4, R=H) as the product of the Michael‐addition of M1 and M2 (R=H) can proceed the Michael‐elimination with a high yield up to ≈95 %, strongly suggesting the Michael‐addition‐elimination as a dynamic chemistry for the C=C bond exchange. Significantly, HT NMR measurements reveal that the addition of water can significantly facilitate the Michael‐addition‐elimination for the C=C bond exchange, reaching equilibrium with a ratio of 1 : 1 for M1 (M3) and M6 (M2) within 5 min. Thereby, the combination of Knoevenagel polycondensation and water‐assisted Michael‐addition‐elimination (KMAE polymerization) is demonstrated as a highly efficient and reproducible strategy for the synthesis of highly crystalline CN‐substituted V‐2D‐COFs. Four examples, including V‐2D‐COF‐TFPB‐PDAN (V‐2D‐COF‐1, also abbreviated as 2D PPV in literature), V‐2D‐COF‐TFPT‐PDAN (V‐2D‐COF‐2), V‐2D‐COF‐TFPB‐BDAN (V‐2D‐COF‐3), and V‐2D‐COF‐HATN‐BDAN (V‐2D‐COF‐4) are demonstrated by using the KMAE polymerization. According to simulated and experimental PXRD patterns and N2‐adsorption‐desorption measurements with matched pore‐size distributions, highly crystalline CN‐substituted V‐2D‐COFs can be synthesized via A3B2 and A6B2 polymerization from various monomers, including C3‐symmetric 1,3,5‐tris‐(4‐formylphenyl)benzene (TFPB), (2,4,6‐tris(4‐formylphenyl)‐1,3,5‐triazine (TFPT), 2,3,8,9,14,15‐hexa(4‐formylphenyl)diquinoxalino[2,3‐a : 2′,3′‐c]phenazine (HATN‐6CHO), and C2‐symmetric (1,4‐phenylene)diacetonitrile (PDAN), and 2,2′‐(biphenyl‐4,4′‐diyl)diacetonitrile (BDAN). Moreover, high‐resolution transmission electron microscopy (HR‐TEM) analysis exhibits crystalline domains with the range of 20–100 nm for V‐2D‐COF‐1, V‐2D‐COF‐3, and V‐2D‐COF‐4.

Figure 1

a) The synthesis of CN‐substituted V‐2D‐COFs via the Knoevenagel and in situ Michael‐Addition‐Elimination polymerization (KMAE polymerization). b) The dynamic Michael‐addition‐elimination for the C=C bond exchange is confirmed by the efficient conversion of diphenylacrylonitrile (M1) and (4‐methylphenyl)‐acetonitrile (M3) to (Z)‐3‐phenyl‐2‐(p‐tolyl)acrylonitrile (M6) and phenylacetonitrile (M2).

Results and Discussion

To testify the synthesis of crystalline frameworks involving a self‐correction process, model reactions for the possible C=C bond exchange are firstly investigated by time‐dependent in situ 1H‐NMR spectroscopy at 95 °C, using N,N‐dimethylformamide‐d7 (DMF‐d7) as solvent, Cs2CO3 as base, and 1,3,5‐trimethoxybenzene as reference to calculate the yields (Figure S1). The model reaction of M1 (1.0 equiv) and M3 (1.0 equiv) generates M2 and M6, indicating a C=C bond exchange process (Figure 2a): The 1H NMR peaks at ≈7.4, ≈7.7, and 2.4 ppm, assigned to the hydrogens (g, h, and f) of compound M2 and hydrogen e and methyl group of M6, respectively, appeared within 5 mins. After a one‐hour reaction time, M6 and M2 were obtained in ≈22 % yield according to the integration of peaks in NMR spectroscopy, with the ratio of ≈7 : 2 for M1 (M3) : M6 (M2).

Figure 2

a) Exploration of the possible C=C bond exchange via anhydrous Cs2CO3 catalyzed model reactions of M1 and M3 into M6 and M2 under anhydrous conditions with the in situ 1H‐NMR study. b) The two possible routes for the C=C bond exchange from compounds M3 and M1 to M6 and M2. Path I: Retro‐Knoevenagel reaction from compound M1 to M2 and M7, plus an additional Knoevenagel condensation between benzaldehyde M7 and M3, which would provide compound M6. Path II: Michael‐addition from compounds M3 and M1 to the intermediate M5, followed by Michael‐elimination from M5 to M6 and M2.

Based on the above NMR investigations, we propose two possible reaction pathways for converting compounds M1 and M3 into M6 and M2 (Figure 2b). Path I involve the retro‐Knoevenagel reaction mechanism from compound M1 to benzonitrile M2 and benzaldehyde M7; afterwards, another Knoevenagel condensation between M3 and M7 would provide compound M6. Path II involves the Michael‐addition of M1 and M3 forming the intermediate compound 1,3‐dicyano‐1,2‐diphenyl‐3‐(p‐tolyl)propan‐1‐ide (M5); the subsequent Michael‐elimination of M5 provides benzonitrile M2 and compound M6.

To determine which reaction pathway is dominant for the conversion (C=C bond exchange) of compounds M1 and M3 into M6 and M2 under anhydrous conditions, the retro‐Knoevenagel reaction of compound M1 into M2 and M7 was also investigated at 95 °C, using DMF‐d7 as solvent, Cs2CO3 as base, and 1,3,5‐trimethoxybenzene as reference to calculate the yields (Figure S2). The time‐dependent 1H‐NMR spectrum demonstrated that only the peaks of compound M1 were identified without observing M2 and M7. This result suggests that the proposed Path I can be excluded. To understand the Path II involving the Michael‐addition‐elimination mechanism, we particularly synthesized the Michael‐addition product M4 (R=H) according to the reported literature (Figure S40–S42). The Michael‐elimination step was performed by subjecting compound M4 at 95 °C, using DMF‐d7 as solvent, Cs2CO3 as base, and 1,3,5‐trimethoxybenzene as reference to calculate the yields using in situ NMR investigation. Remarkably, after one‐hour reaction, all 1H‐NMR ‐peaks in the range of 3.5–5.5 ppm belonging to the aliphatic hydrogens i and j of M4 (Figure 2b, R=H) disappeared (Figure S3). Instead, new NMR signals at ≈7.4 and ≈7.8 ppm were observed, which are assigned to the hydrogens (g, h, and f) in M2 and hydrogens c or d in M1 (R=H). The corresponding yield of compound M2 and M1 (Figure 2b, R=H) is estimated to be ≈95 %. Thereby, we can conclude that Path II (Michael‐addition‐elimination) is rational for the C=C bond exchange, which is distinguished from the hypothetical reversible Knoevenagel reaction for the first time.[ , ]

Encouraged by the above results, we further propose a four‐step mechanism for the Michael‐addition‐elimination reaction by using Cs2CO3 as a catalyst (Figure 3a). The first two steps belong to the Michael‐addition. In Step 1, the carbon anion specie M3a nucleophilic‐ally attacks the CN‐substituted vinylene linkage of compound M1, providing the intermediate anion M9. In Step 2, the intermediate M9 is converted into the neutral intermediate M5 via protonation. The next two steps deal with the Michael‐elimination: In step 3, the deprotonation of M5 yields anion specie M10. Then in step 4, M10 dissociates into M2a and M6. Overall, the conversion is M3a+M1<=>M2a+M6.

Figure 3

a) Proposed mechanism for the Michael‐addition‐elimination reaction. b) 1H‐NMR study of aqueous 0.1 M Cs2CO3 catalyzed substitution reaction between M3 and M1 for the C=C bond exchange.

In the proposed mechanism, both Steps 2 and 3 (Figure 3a) involve proton‐transfer processes. Under Cs2CO3‐catalyzed conditions, the HCO3 − and CO3 2− can be used as a donor/acceptor for the proton transfer. Consequently, the Michael‐addition‐elimination occurs in a slow rate due to the relevantly limited amount of proton. In this respect, we assume that the Michael‐addition‐elimination for the C=C bond exchange can be mostly accelerated by the addition of small amount of water, which would generate sufficient H2O/OH− as proton‐donor/acceptor pair to replace the HCO3 −/CO3 2− pair. It should be noted that the softness of Cs+ makes Cs2CO3 rather soluble in organic solvents such as DMF and alcohol, which renders Cs2CO3 an excellent candidate base for studying the effect of water.[ , ]

To examine the influence of water on the Michael‐addition‐elimination, as shown in Figure 3b, the reaction between M3 (1.0 equiv) and M1 (1.0 equiv) was further studied by time‐dependent in situ 1H‐NMR at 95 °C with DMF‐d7 as the solvent and aqueous 0.1 M Cs2CO3 solution as the catalyst. Remarkably, the peaks at ≈7.8 (for M1) and ≈7.7 ppm (for M6) of the 1H‐NMR spectrum suggested a ratio of 1 : 1 for M1 : M6 within 5 mins. Afterwards, the equilibrium was maintained (Figure S4). In contrast, the Michael‐addition‐elimination reaction with anhydrous condition presented a ratio of 7 : 2 for compounds M1 : M6 even after 1 h reaction time. This sharp contrast clearly demonstrates the crucial role of water in accelerating the C=C bond exchange. The Michael‐elimination of intermediate M4 was also investigated under the water‐assisted condition, which resulted in compounds M1 and M2 with a yield of ≈92 % within 5 mins (Figure S6). It should be noted that the addition of water also enables the retro‐Knoevenagel reaction of M1 with a low yield for M2 and M7 (<10 %) in a one‐hour reaction (Figure S7), which can additionally contribute to the efficiency of C=C bond exchange (Figure 2b).

To gain a deeper insight into the Michael‐addition‐elimination for the C=C bond exchange, DFT calculations were conducted to calculate the energy of each intermediate of the Knoevenagel condensation between M2 and M7 (Figure S8, S12–S14), and the Michael‐addition‐elimination between M3 and M1 (Figure 4a, Figure S9–S11) with zero, four and eight water molecules. For the Michael‐addition‐elimination, the intermediates of cesium 1,3‐dicyano‐2,3‐diphenyl‐1‐(p‐tolyl)propan‐1‐ide M8 and M10 possess the highest energy levels, which should determine the reaction rates. The energies of M8 and M10 are decreased by 0.12 and 0.33 eV with eight surrounding water molecules, compared to the corresponding ones without hydration (Figure 4b). This result further indicates that the addition of water can facilitate the Michael‐addition‐elimination reaction. It should be noted that the much lower energy barrier of the Michael‐addition‐elimination (EBMAE=0.12 eV, Figure 4b) than the Knoevenagel (EBk=0.43 eV, Figure S8) and retro‐Knoevenagel (EBrk=1.02 eV, Figure S8) reaction also suggests that the Michael‐addition‐elimination is the main driving force for the achieved C=C bonds exchange, which matches very well with above model reaction results (Figure 3b).

Figure 4

a) The Michael‐addition‐elimination route from M3a and M1 to M2a and M6. b) DFT calculation of Michael‐addition‐elimination reaction without water (black), with four water molecules (blue), and with eight water molecules (red).

Encouraged by the dynamic covalent nature of the Michael‐addition‐elimination for the C=C bond exchange as described above, the combination of Knoevenagel polycondensation and water‐assisted dynamic Michael‐addition‐elimination (classified as KMAE polymerization) was further demonstrated for the 2D polycondensation. The firstly reported CN‐substituted V‐2D‐COF (V‐2D‐COF‐1, Figure 5a) can be synthesized by heating 1,3,5‐tris‐(4‐fromylphenyl)benzene (TFPB, 1.0 equiv) and (1,4‐phenylene)diacetonitrile (PDAN,1.5 equiv) in the mixture of N,N‐dimethylacetamide (DMAc)/Cs2CO3 (0.1 M) at 120 °C for 3 days. After washing and drying, V‐2D‐COF‐1 was obtained as yellowish powder in 80 % yield.

Figure 5

a) The synthesis of V‐2D‐COF‐1 via the water‐assisted KMAE polymerization conditions. b) Corresponding experimental and simulated PXRD patterns of V‐2D‐COF‐1. c) HR‐TEM image of V‐2D‐COF‐1 synthesized via water‐assisted condition. d) The comparison of PXRD patterns of V‐2D‐COF‐1 synthesized with and without the addition of water (measure parameters: step size: 1°, time: 60 s). e) The comparison of PXRD patterns of different V‐2D‐COF‐1 batches synthesized via the water‐assisted conditions.

The simulated and experimental PXRD patterns as well as the difference plot confirmed the formation of crystalline frameworks (Figure 5b). Moreover, the high‐resolution transmission electron microscopy (HR‐TEM) study was carried out for V‐2D‐COF‐1. The HR‐TEM images revealed crystalline domains with periodic patterns of ≈2.96 nm, which matched well with the theoretical d100‐spacing value of ≈3.0 nm (Figure 5c). In contrast, the PXRD patterns of V‐2D‐COF‐1 synthesized under strictly anhydrous reaction conditions showed much lower crystallinity (Figure 5d). Using the same PXRD parameters, the intensity of the (100) reflex (≈6000) of V‐2D‐COF‐1 synthesized under water‐assisted conditions is ≈3 times higher than the value (≈2000) of V‐2D‐COF‐1 that was synthesized without the addition of water. The lower crystallinity of V‐2D‐COF‐1 synthesized under the anhydrous condition can be explained by insufficient the error‐correction process without involving the efficient C=C bond exchange. To reveal the reproducibility of the water‐assisted condition, the synthesis of V‐2D‐COF‐1 under aqueous 0.1 M Cs2CO3 catalyzed polymerization conditions were further conducted randomly five times. The PXRD patterns of all five batches from V‐2D‐COF‐1 showed superior reproducibility (Figure 5e).

To demonstrate the versatility of the KMAE polymerization for the synthesis of various CN‐substituted V‐2D‐COFs, another three examples of the CN‐substituted V‐2D‐COFs (Figure 6a), including V‐2D‐COF‐TFPT‐PDAN (V‐2D‐COF‐2), V‐2D‐COF‐TFPB‐BDAN (V‐2D‐COF‐3), and V‐2D‐COF‐HATN‐BDAN (V‐2D‐COF‐4) from different aromatic aldehydes [C3‐symmetric 1,3,5‐tris‐(4‐fromylphenyl)benzene (TFPB), (2,4,6‐tris(4‐formylphenyl)‐1,3,5‐triazine (TFPT), 2,3,8,9,14,15‐hexa(4‐formylphenyl)[2,3‐a : 2′,3′‐c]phenazine (HATN‐6CHO)) and aromatic nitrile [C2‐symmetric (1,4‐phenylene)diacetonitrile (PDAN) and 2,2′‐(biphenyl‐4,4′‐diyl)diacetonitrile (BDAN)] were also synthesized successfully under aqueous 0.1 M Cs2CO3 catalyzed conditions. Experimental and simulated PXRD patterns revealed that all four V‐2D‐COFs are highly crystalline. The detailed synthesis and characterization were described in the Supporting Information (Figure S15–S29).

Figure 6

a) The chemical structures of V‐2D‐COF‐2, V‐2D‐COF‐3, and V‐2D‐COF‐4 that newly synthesized from several monomers, including TFPB, TFPT, HATN‐6CHO, PDAN, and BDAN. b) N2 adsorption–desorption isotherms and c) pore size distributions of V‐2D‐COFs synthesized via water‐assisted KMAE polymerization conditions; HR‐TEM images of d) V‐2D‐COF‐3 and e) V‐2D‐COF‐4, which are synthesized via water‐assisted conditions.

N2 adsorption‐desorption measurements were performed to investigate the porosity of the above‐mentioned V‐2D‐COFs. The BET surface areas of V‐2D‐COF‐1, V‐2D‐COF‐2, V‐2D‐COF‐3, and V‐2D‐COF‐4 are 603, 287, 127, and 504 m2 g−1, respectively (Figure 6b). Further nonlocal density functional theory (NLDFT) calculations reveal the pore size distributions of 2.3–3.2 nm for V‐2D‐COF‐1 and V‐2D‐COF‐2, 3.1–4.2 nm for V‐2D‐COF‐3, ≈1.3 and ≈1.8 nm for dual‐pore V‐2D‐COF‐4 (Figure 6c). Notably, the surface areas of V‐2D‐COF‐1 (602 m2 g−1) and V‐2D‐COF‐2 (287 m2 g−1) are higher than the values of corresponding reported V‐2D‐COFs synthesized via anhydrous Cs2CO3‐catalyzed conditions (472, and 232 m2 g−1).[ , ] It should be mentioned that these values are still lower than the theoretic surface areas (≈2133.4 m2 g−1 for V‐2D‐COF‐1, ≈2013 m2 g−1 for V‐2D‐COF‐2, ≈2181.9 m2 g−1 for V‐2D‐COF‐3, ≈1376 m2 g−1 for V‐2D‐COF‐4). High‐resolution transmission electron microscopy (HR‐TEM) images of both V‐2D‐COF‐3 (Figure 6d) and V‐2D‐COF‐4 (Figure 6e) powder samples displayed polycrystalline structures, and the crystalline domain sizes range from 20 to 100 nm with periodic patterns of ≈3.64, and ≈2.8 nm, which matched well with the theoretical d100‐spacing values of 3.6, and 2.8 nm, respectively.

Conclusion

In conclusion, we demonstrate a novel method for synthesizing highly crystalline V‐2D‐COFs via the combination of Knoevenagel polycondensation and water‐assisted Michael‐addition‐elimination (KMAE polymerization). Through in situ high‐temperature NMR measurements of model reactions and intermediate, the Michael‐addition‐elimination is clearly distinguished from the hypothetical reversible Knoevenagel reaction for the first time. Furthermore, the water‐assisted Michael‐addition‐elimination is assigned as an efficient dynamic covalent chemistry for the direct C=C bond exchange, which endows the self‐correction property to synthesize complicated extended structures. Indeed, DFT calculations indicate a reduced energy barrier of Michael‐addition‐elimination by the addition of water. The deep understanding of the reaction mechanism further guides the synthesis of four highly crystalline V‐2D‐COFs with crystalline domain sizes ranging from 20 to 100 nm. This work not only provides efficient dynamic C=C exchange chemistry for the synthesis of highly crystalline V‐2D‐COFs, but also pave the way for the future development of single crystalline V‐2D‐COFs and new type of vinylene‐linked 2D‐conjugated polymers for optoelectronic applications.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Click here for additional data file.

Supporting Information

Click here for additional data file.

Supporting Information

Click here for additional data file.

Supporting Information

Click here for additional data file.

Supporting Information

Click here for additional data file.

Supporting Information

Click here for additional data file.

Supporting Information

Click here for additional data file.

Supporting Information

Click here for additional data file.