Assembly of metal acceptors with appropriate organic ligands through dynamic coordination bond has become an effective technique for building discrete cage-like molecules of desired shapes and sizes. However, assemblies formed by coordination bond coordination are generally particularly sensitive to heat, acids and alkalis due to the instability of metal coordination bonds. However, due to the irreversibility of the amide bond formation reaction, many polymers are easily formed during the amide bond process rather than the formation of discrete organic cages. Therefore, building discrete organic cages from multiple kinetic inert bonds (such as amide bonds) is very challenging and often requires multiple steps of reaction. Due to the lack of a self-correction process, kinetic controlled amide bond formation reactions often produce mixed products. Thus, dynamic covalent chemistry (DCC) driven by the imine condensation reaction allows chemists to obtain many porous organic cages of different shapes and sizes at high yields through a one-pot reaction. Despite its high yields in formation, the use of imine cages is limited because imine bonds are prone to hydrolyzing in acidic, basic and aqueous media. On the contrary, amide bonds have good chemical stability in nature. In the process of oxidizing imine to amide bonds, the Pinnick reaction provides good yields under mild reaction conditions. Therefore, modifying imine analogs to amide analogs after synthesis is a potential way to build a long-lasting and robust amide -based structure. In addition, the multiple amide bonds present in the cavity of the amide cage are good hydrogen bond donors, so that the amide receptor can act as a good potential anion receptor.
Based on the above research background, the author uses dynamic imine chemistry to obtain the [2+3] imine cage by condensation of diamine (A) and triacetaldehyde (P), and then oxidize the unstable imine bonds into amide bonds, thereby forming a solid cage C2 containing multiple amide bonds (Scheme 1). The universality of the method is further expanded by forming an imine macrocycle (M1) and synthesizing an amide macrocycle (M2) containing four amide bonds by post-synthesis (Scheme 2). The crystal structure of macrocycle M2 reveals the hydrogen bond between the amide bond and the solvent molecule DMSO, resulting in the rapid formation of macrocycle M2 aggregates, limiting its application in anion binding. However, amide C2 does not form any aggregates in solution due to its inherent 3D structure. Cage C2 has chemical and thermodynamic stability, and its closed cavity containing six amide bonds shows high selectivity to F- with a binding ratio of 1:2. The binding affinity of C2 to F- is about 3×103 times that of Cl-, but has no effect on other halides , phosphate and oxygen anions.
Scheme 1 Scheme 1 Scheme 1 Synthesis of the amide cage (C2) by oxidation after synthesis of imine analog (C1).
is different from dynamic covalent chemistry. Kinetic controlled amide bond formation reactions usually form multiple products due to the lack of a self-correction process. For example, using the traditional method, reacting terephthalyl chloride with diamine (A) in DMSO at a molar ratio of 1:1 will result in a variety of products without forming a simple [2+2]amide macrocycle (Scheme S1). In this regard, post-synthesis oxidation of imine macrocycles is another easy way to obtain amide macrocycles in a single step (Scheme 2).
Scheme 2 Synthesis route for the conversion of imine macrocycle (M1) to amide macrocycle (M2).
The authors synthesized imine macrocycle (M1) using an optimization scheme and characterized the resulting macrocycle by 1H and 13C nuclear magnetic resonance and infrared spectroscopy. For post-synthesis oxidation, macrocyclic M1 was dissolved in THF and treated with the mild oxidant NaClO2 in the presence of 2-methyl-2-butene. Next, a weakly acidic NaH2PO4 aqueous solution was added dropwise to the reaction mixture, and stirred at room temperature for 20 h to obtain a colorless solution, which was concentrated and treated with a mixture of water and acetonitrile to obtain a milky white precipitate. Subsequently, the product was separated into a white precipitate by centrifugation (50% yield) and washed with a mixture of chloroform and acetonitrile. The conversion of imine to amide functional groups was confirmed by infrared spectroscopy: the stretching frequency of C=N (1630cm-1) decreased, and a new peak of the stretching frequency of amide NH (3314cm-1) and carbonyl C=O (1662cm-1) was generated (Figures S10, S12). Surprisingly, when the solid dissolves in DMSO, the solution is immediately cloudy.The NMR spectrum of the turbid solution (DMSO-d6) shows that there are many broad peaks of low intensity due to the formation of white precipitates (Fig. S6). THF was added to the cloudy DMSO solution of the macrocyclic ring, and the solution was clarified again, indicating that aggregates were formed in DMSO. Therefore, M2 was immediately subjected to NMR studies after dissolving in DMSO-d6/THF-d8, and a clear NMR spectrum was obtained (Figures S7 and S9): M2 had a new peak at 9.97 ppm and there was no imine peak, which again indicated that the imine bond was completely converted into an amide bond.
Then, it can be seen from the single crystal structure of the [2+2]amide macrocycle (M2) (Figure 1): In the crystal structure, the asymmetric unit contains a complete macrocycle molecule (M2) and a DMSO and a THF molecule. The average distance between the two terminal parallel benzene rings is 21.0, while the center distance between the two non-parallel central benzene rings is about 5.2. The O atom of DMSO and the amide NH of macrocycle M2 are within the hydrogen bonding range, so there is a strong hydrogen bond between M2 and DMSO, where the distances of N···O and NH···O are 2.9 and 2.0, respectively. In addition, the single crystal structure also shows that strong hydrogen bonds promote intermolecular accumulation of M2 units to form aggregates, and there is no interaction between THF and M2. Therefore, M2 easily forms aggregates in DMSO, limiting its application as anion-binding receptor in solution state. In this case, the authors considered designing the amide cage, hoping that it would not work in aggregation formation, as there is a lack of effective molecular filling due to the inherent 3D arrangement. Therefore, the authors selected trialdehyde (P) and diamine (A) to condensate to give imine cage (C1) (Scheme 1) and characterized by 1H NMR, 13C NMR, 1H-13C HMQC, 2D DOSY NMR, FTIR spectroscopy and ESI-HRMS. The nuclear magneto-spectrum of imine cage C1 showed that a new peak was produced at 8.32 ppm ((-HC=N)) and the aldehyde peak at 10.10 ppm disappeared (Fig. 2b). The composition of the cage was accurately determined through ESI-HRMS analysis: significant signal peaks of [M+H]+, [M+Na]+, [M+K]+ and [M+2H]2+ were shown at 1706.8805 (calculated value 1706.8707), 1728.8474 (calculated value 1728.8532), 1744.8214 (calculated value 1744.8272) and 853.9434 (calculated value 853.9390), respectively, demonstrating the formation of the [2+3]imine cage (C1). The optimized structure of DFT shows that the imine cage C1 is a triangular prism structure, with the spacing between the two central benzene rings of the triangle surface and the length of each arm are 5.3 and 13.7 respectively.
Figure 1 (a) Single crystal structure of M2. (b) Crystal stacking view along the crystal c-axis.
Next, the imine cage C1 was subjected to Pinnick oxidation according to the same synthesis scheme as amide macrocycle M2 (Scheme 1), and the obtained amide cage (C2) was a white solid with a yield of 48%. Nuclear magnetic characterization was initially performed in DMSO-d6 (Fig. S26), which did not form any aggregates in solution (Fig. 2c): the characteristic imine formant peak completely disappeared and a new peak was produced at 9.85 ppm (amide), indicating that the imine bond was completely converted to an amide bond. The 13C nuclear magnetic spectrum showed that a new peak appeared at 164.52 ppm, which had no correlation with amide protons (9.85 ppm), indicating the formation of amide-bonded carbon atoms instead of imine carbon atoms (Figures S30, S31). Infrared spectroscopy also confirmed the conversion of imine to amide functional groups: a C=N (1629cm-1) stretching vibration peak was generated, and a amide NH (3291 cm-1) and carbonyl C=O (1655cm-1) stretching vibration peak was generated (Figures S32 and S33). Finally, the ESI-HRMS mass spectrometry had significant peaks of [M+Na]+ and [M+2Na]2+, further supporting the composition of the amide cage (C2) (Fig. 3).
Figure 2 Partial NMR spectrum: (a) Trialdehyde (P) (CDCl3), (b) imine cage (C1) (CDCl3), (c) amide cage (C2) (DMSO-d6).
Figure 3 ESI-HRMS spectrum of C2: (b) Simulation and (c) Experimental isotope distribution pattern.
DFT optimized structure of amide cage C2 can be seen (Figure 8a): its structure is two propeller-shaped panels composed of three amide clips. The benzene ring and amide bonds of the two triangular plates are almost parallel to each other. The distance between the two opposite central benzene rings is 5.0. The distance between relative N atoms in the triple pair amide bond is in the range of 4.8-5.3. This arrangement of amide functional groups produces three diamide bags and is well separated laterally, so that C2 can serve as a good receptor with multiple binding sites.In addition, after adding strong acid or strong alkali, the amide cage C2 still has good chemical stability; the thermal stability of amide cage C2 is also as high as 523℃.
excellent chemical/thermal stability and solubility enable amide cage (C2) to become a potential anionic receptor. The presence of three blocked diamide capsules of suitable sizes with hydrogen bond donor units is expected to make them selective for binding of fluoride . Therefore, by 1H NMR analysis (Fig. S38), the binding properties of cage (C2) to different anions (NO3-, ClO4-, PF6-, I-, Br-, Cl-, F-) TBA salts in polar media (such as DMSO-d6). After adding excess (10 equivalents) of TBA salt, the nuclear magnetic changes of the main body (C2) were detected. The results showed (Figure 4): For NO3-, ClO4-, and PF6-, due to the large anion size, it was impossible to enter the restricted diamide pocket of the cage. In the halogen ions, there is no significant change in the nuclear magnetism of cage C2 for I-. For Br-, the displacement of amide NH (Ha) protons of cage C2 is negligible (Δδ=0.03 ppm). For smaller Cl-, the amide proton NH (Δδ = 0.19 ppm) of cage C2 and the associated aromatic protons are slightly shifted. Therefore, cage C2 has a certain binding affinity for Cl-. For F-, the authors added 10 equivalents of TBAF solution (1M THF solution) to the DMSO-d6 solution of cage C2, and as expected, both the amide proton NH (Δδ = 4.9 ppm) and the phenyl proton CH (Hb) of cage C2 were moved significantly towards the low field. This result shows that the amide cage C2 has a strong binding ability to F-, which is attributed to the comparison of the size of the diamide pocket of C2 to F-. Afterwards, the binding degree of amide cage C2 to F- was further studied by nuclear magnetic titration. With the addition of TBAF, amide NH (Ha) gradually widens and moves toward a low field, and the associated phenyl CH (Hb) protons also significantly move out of the field.
Figure 4 Stacked nuclear magnetic spectrum (DMSO-d6): (a) amide cage (C2), and 10 equivalents of TBA salts (b)I-, (c)Br-, (d)Cl- and (e)F- were added to C2, respectively.
Figure 5 Nuclear magnetic titration spectrum of amide cage C2 (DMSO-d6).
Then, the binding constant was calculated by data fitting of the nuclear magnetic titration through bind-fit (Fig. 7). For F-, the experimental titration data were more consistent with the 1:2 (mass-guest ratio) binding model. From the fitting curve, it can be seen that the secondary binding constant K2=6×104M-1 of cage C2 and F- is very high, while the primary binding constant (K1=0.6 M-1) with F- is very low, which indicates that only a 1:2 host-guest complex is formed, not a 1:1 complex. For Cl-, the experimental data fit better with the 1:1 (mass-to-guest ratio) model, and the binding constant K1=19.7 M-1 is very low. Therefore, the binding strength of the organic cage C2 to F- is 3×103 times that of Cl-, and there is no binding to other anions.
Finally, the author explored the reason why cage C2 selectively binds F- through DFT theoretical calculations (Figure 7): free cage C2 has three diamide pockets, and the two N atoms in each pocket are 4.8-5.3 apart. In the host-guest optimized structure, the average NH···Cl distance of Cl- located outside the diamide capsule is 2.4. Optimization using two Cl- will cause a larger displacement of the second Cl-from the binding bag (NH··Cl distance 2.5), with only one NH···Cl hydrogen bond, indicating an effective binding of 1:1 (Figure S49). On the other hand, in the subject-guest energy-optimized structure (2F·C2), the two F- are well wrapped in the two diamide pockets of cage C2 (Fig. 8b). However, the optimization of the three F-binding causes the third F- to be displaced from the center of the third diamide bag (NH····F bond distance 1.6) (Fig. S48). Therefore, the density functional theory study explains the reason why fluoride ions are highly selectively bound by only a 1:2 host-guest stoichiometric ratio.
Figure 6 Draw the nuclear magnetic titration curve of amide cage C2 based on the chemical shift changes of (a) amide proton (Ha) and (b) phenyl proton (Hb) to guest equivalent (TBAF/TBACl).
Figure 7 (a) Ball model of cage C2 energy optimization structure. (b) The stick model of the subject and object optimizes the structure of 2F·C2.
In summary, the authors reported a strategy using dynamic imine chemistry, which condensate with dialdehyde and trialdehyde, respectively, to obtain the [2+2]imine macrocycle and [2+3]imine cage. After synthesis, the imine bond is converted into amide bonds through Pinnick oxidation to form the required [2+2]amide macrocycle and [2+3]amide cage. Among them, amide cage C2 is stable against nitric acid, sodium hydroxide and heat (~523°C). The study found that a solid cage C2 with a confined space surrounded by multiple amide parts has good selective recognition performance for F-, and its binding constant is about 3×103 times stronger than Cl-. Interestingly, cage C2 does not recognize other halogens, phosphates and oxygen anions. Future work will focus on expanding the range of robust covalent structures with structural diversity for potential applications of post-synthesis conversion using dynamic imine bonds to kinetic inert moieties that would otherwise be difficult to achieve by conventional synthesis strategies.
Title: Post-Synthesis Conversion of an Unstable Imine Cage to a Stable Cage with Amide Moieties towards Selective Receptor for Fluoride
Author: Pallab Bhandari, and Partha Sarathi Mukherjee*
DOI: 10.1002/chem.202201901
