Metal-organic cages (MOCs) with diverse nanocavities created through coordination directed self-assembly have been the subject of thorough research due to their unique characteristics [
1–
9]. These cages are capable of accommodating a range of guest species, making them valuable for tasks such as chemical separations [
10–
12], stable active species [
13–
15], catalysis [
16–
18], and drug delivery [
19,
20]. In order to achieve the above functions, it is crucial and challenging to cleverly utilize supramolecular interactions and spatial effects that affect the binding ability between host and guest species, such as electrostatic affinity, hydrogen bonding,
π-stacking, solvent effect, and steric hindrance [
21–
23].
The ligand modification in MOCs is a common approach involving replacing specific ligand substituents in a rational manner, exploiting electronic or spatial effects to directly influence the structure and properties of the cages [
24–
28]. Previously reported examples show that MOC structures with different topological or symmetrical types can be constructed by appropriately replacing substituents [
29–
34]. For instance, Fujita
et al. created a giant M
30L
60 icosidodecahedron by adjusting the flexibility of the ligand through modifications with different numbers of methyl groups [
29]. Nitschke
et al. reported the successful construction of two new Cu
12IL
8 cages with different structures using similar subcomponents except for the substituents (
e.g., –H, –F and –OCH
3) in the components [
34]. However, the predictability of MOC topological structure changes caused by local substituent modification is unclear, which poses a challenge for studying the structure-activity relationships of its subsequent host-guest chemistry. Some interesting studies have shown that by appropriate vertex modification while keeping the MOC topology unchanged, the host-guest properties of the cage cavity can be well controlled [
35–
39]. For example, the subcomponent exchange converted the Fe
Ⅱ centers of a tetrahedral cage from high-spin to low-spin state by replacing 2-formyl-6-methylpyridine with 2-formylpyridine, endowing the cage system with the ability to responsively release guests [
35]. The development of such cage systems in which topology is maintained after vertex modifications will provide an excellent platform for studying attractive host-guest chemistry.
Here, we explore a vertex functionalization method to study the influences of the host-guest chemistry of MOCs by comparing fluorinated hexahedral MOC
1 and non-fluorinated MOC
2 [
40,
41]. Structural characterization
via single-crystal X-ray diffraction (SCXRD),
1H nuclear magnetic resonance (NMR) spectra, electrospray ionization time-of-flight mass spectra (ESI-TOF MS) confirmed that MOC
1 adopts low-symmetry architectures similar to MOC
2, with fluorine atoms positioned at cage windows.
1H/
19F NMR spectra of
1·NTf2 and
1·OTf exhibited slow-exchange binding behavior on the NMR time scale with the external anions. The
1H NMR spectrum of
1·NTf2 showed a single set of peaks in the imine signal area, while
1·OTf displayed two distinct groups of signals. This suggests that the NTf
2- anion has a stronger binding affinity with the cage. Competitive anion titrations also demonstrated a hierarchy in binding strength (NTf
2- > OTf
- > BF
4-), with NTf
2- resisting displacement by smaller anions or aromatic guests. Host-guest studies showed pyrene fully displaced OTf
- but not NTf
2-, underscoring the superior affinity of NTf
2-. Isothermal titration calorimetry (ITC) further revealed reduced association constants for
1·OTf compared to
2·OTf when binding polycyclic hydrocarbons. This was attributed to the electron-withdrawing fluorine effects altering charge distribution. These findings highlight how vertex modification fine-tunes anion selectivity and guest-binding thermodynamics in MOCs, offering a strategic approach to designing adaptive supramolecular hosts for applications.
The reaction of 1,2,4,5-tetrakis-(4-aminolphenyl)benzene (TAPB), 3-fluoro-2-formylpyridine, and ZnX
2 (X including NTf
2-, OTf
-, and BF
4-) salt in acetonitrile (CH
3CN) at 70 ℃ for 8 h gave MOC
1 (
Fig. 1, see Supporting information for synthesis experimental details). The products corresponding to different counteracting anions were named as
1·NTf2, 1·OTf, and
1·BF4. As comparative structures with the same geometric shape, the previously reported MOC
2 (marked as
2·OTf, 2·NTf2, and
2·BF4) [
40,
41] was synthesized by replacing 3-fluoro-2-formylpyridine with 2-formylpyridine in the former synthesis conditions. The
1H NMR spectra of MOCs
1 and
2 were recorded and showed in
Fig. 2, Figs. S3, S13, S25, and S35-S37 (Supporting information). The
1H NMR spectrum of
1·NTf2 in CD
3CN displayed four distinct signals in the imine region with integrated area ratio of 1:1:1:1 (H
1,
Fig. 2), which are similar to the signals of
2·NTf2, indicating that they may have similar low-symmetry cage structures [
41]. However, it is evident that the
1H NMR signal distribution of
1·OTf was significantly different from that of
2·OTf. Eight signals appeared in the imine region, and their integrated area ratios were 1:1:1:1:1:1:1:1 (
Fig. 2), suggesting that they might belong to two different forms of cage species in the solution. The similar results appeared in the
1H NMR spectra of
1·BF4 after comparison with
2·BF4 (
Fig. 2). The above phenomenon prompted us to speculate that the modification of the cage vertices resulted in a significant change in the host-guest binding behavior between the cage and the anions.
In order to understand the reasons for the differences in the above
1H NMR data,
1·NTf2 and
1·OTf were selected as key research objects for detailed characterization. The diffusion ordered
1H NMR spectrum (DOSY) and ESI-TOF MS were firstly measured to investigate the composition of
1·NTf2. As showed in Fig. S4 (Supporting information), all the peaks of
1·NTf2 had the same diffusion constant (
D = 5.50 × 10
-10 m
2/s), indicating that these peaks belonged to a single species. The hydrodynamic radius of
1·NTf2 in CH
3CN was calculated to be ~11.9 Å
via the Stokes-Einstein equation [
42,
43]. At the same time, ESI-TOF mass spectra (Figs. S5 and S6 in Supporting information) showed that
1·NTf2 had multi-charge features and the compositional characteristic of [Zn
8L
6(NTf
2)
16-n]
n+ (
n = 10–4). The main peak observed at
m/
z = 1181.0524 matched the [
1·(NTf
2)
9]
7+ species (calculated value 1180.9854). Additionally, all peaks corresponding to the different charged species aligned well with their calculated isotopic patterns, confirming the composition and stability of
1·NTf2 in the CH
3CN solution. In addition, purple block crystals of
1·NTf2 that suitable for single crystal X-ray diffraction (SCXRD) analysis were obtained by diffusing ethyl acetate vapor into its CH
3CN solution containing iodine as crystallization-aiding molecules (see Supporting information for detailed crystallographic data).
1·NTf2 crystallizes in the monoclinic space group
P2
1/
c, and the cage molecule features an 8-nucleus hexahedral structure (
Fig. 3a). Six TAPB-based ligands act as faces of the cage with an average distance of about 12.25 Å between the face centers, bridging the eight octahedral Zn
Ⅱ centers and forming ΔΔΛΛΛΛΛΛ or ΛΛΔΔΔΔΔΔ stereo configurations, which are similar to that of
2·NTf2 (
Fig. 3b) [
41]. The hydrophobic cavities of
1·NTf2 and
2·NTf2 have the volume of 492 Å
3 and 482 Å
3, respectively, estimated by MoloVol with a probe radius of 1.2 Å [
41]. It is worth noting that the modified fluorine atoms on the ligand surround the vertex and are arranged at the window of
1·NTf2 (
Fig. 3). These structural features may have an impact on its host-guest behavior.
Subsequently, a variety of 2D NMR spectra were utilized to validate the structure of
1·NTf2 in solution (Figs. S8-S10 in Supporting information). The heteronuclear single quantum correlation (HSQC) spectrum of
1·NTf2 (Fig. S8) showed that there was one set of imine signals belonging to the cage species in the imine region. Moreover, the
19F NMR spectrum of
1·NTf2 was recorded and showed two sets of signal distribution (Fig. S11 in Supporting information). From the region of −119 ~ −122 ppm, there were four different F signals attributed to pyridine parts, and their integrated area ratio was 1:1:1:1. It was consistent with the situation of imine signals, further proving that
1·NTf2 had a low symmetry structure. Another group of signals came from NTf
2- anions. It was found that the peaks at −70.4 and −80.1 ppm corresponded to encapsulated and free NTf
2- anions respectively [
44]. Based on the above data, it can be concluded that there is a host-guest binding between the cage and the NTf
2- anion, and this process exhibits slow exchange on the nuclear magnetic time scale [
34,
44–
46] without affected by temperature (Fig. S12 in Supporting information).
Similar characterizations were performed on
1·OTf to explore its different host-guest binding behaviors. The DOSY
1H NMR spectrum and ESI-TOF MS of
1·OTf were tested and showed in Fig. S14 (Supporting information). All the peaks of
1·OTf had the same diffusion constant (
D = 3.54 × 10
-10 m
2/s). At the same time, ESI-TOF mass spectra (Figs. S15 and S16 in Supporting information) of
1·OTf had multi-charge species and the compositional characteristic of [Zn
8L
6(OTf)
16-n]
n+ (
n = 9–4). The main peak observed at
m/
z = 1012.8159 matched the [
1·(OTf)
9]
7+ species (calculated value 1012.8835). All signals attributed to the various charged species matched their calculated isotopic patterns. Furthermore, several 2D NMR spectra were examined (Figs. S18-S20 in Supporting information), including the HSQC spectrum of
1·OTf (Fig. S18), which revealed two distinct sets of imine signals corresponding to different cage species in the imine region. The
19F NMR spectrum of
1·OTf was shown in Fig. S21 (Supporting information). The presence of two groups of signals in the range of −119 ~ −122 ppm also supported the inference that there were two different species in the solution. The signal at −78.25 ppm indicated that some of OTf
- anions were encapsulated in the cage. Additionally, the
1H NMR spectra of
1·OTf with varying concentrations and temperatures (Figs. S22-S24 in Supporting information) indicated that the composition of species in the solution was minimally influenced by these factors. More importantly, the crystal structure of
1·OTf (see Supporting information for detailed crystallographic data) shows that it has a similar structure to
1·NTf2, indicating that the differences between the two species mentioned above are caused by the host-guest interactions between the cage and anions, rather than the symmetry or a new structure of the cage. Therefore, we speculate that
1·OTf has the same cage skeleton as
1·NTf2, but the complex
1H NMR spectrum (
Fig. 2) should be caused by the different degrees of slow exchange process between OTf
- and the cage on the NMR time scale [
44–
46]. The binding behavior of
1·OTf is different from
1·NTf2, 2·OTf, and
2·NTf2, indicating that through vertex modification, MOC
1 has acquired the ability to selectively bind with anionic guests.
To further investigate the changes in binding ability between anionic guests and metal-organic cages caused by vertex modifications, we performed a series of experiments involving anion titrations. The general procedure involves dissolving
1·NTf2 or
1·OTf in CD
3CN (2.5 × 10
-3 mol/L), followed by the addition of 16 equiv. of tetrabutylammonium trifluoromethanesulfonate (TBAOTf) or tetrabutylammonium bis((trifluoromethyl)sulfonyl)amide (TBANTf
2), respectively. The solutions were equilibrated at 70 ℃ for 3 h, and then the system was characterized by
1H NMR and
19F NMR. As shown in
Fig. 4a and Fig. S40 (Supporting information), with the addition of 16 equiv. of OTf
- anion, there is no change in the aromatic region of the
1H NMR spectra of
1·NTf2. Additionally, the
19F NMR spectra of the system (marked as
TBAOTf@1·NTf2,
Fig. 4b) indicate that in the range of −79 ~ −80 ppm, upon the addition of OTf
− anion, the bound NTf
2- anions remain unchanged and there are no signals indicating the presence of encapsulated OTf
- anion. Furthermore, the chemical environment of the fluorine atoms on the cage remains consistent (Fig. S41 in Supporting information). These initial findings suggest that the MOC
1 has a higher binding affinity for NTf
2- anions compared to OTf
- anions. To further validate this result, the
1·OTf solution was titrated with NTf
2- anions. As shown in
Figs. 4c and d, the addition of NTf
2- anions has resulted in significant changes in the
1H and
19F NMR spectra of the system (marked as
TBANTf2@1·OTf). With the addition of NTf
2- anions, the original two sets of signals of
1·OTf (
Fig. 4c) gradually merged into a single group, and at the endpoint of titration, the
1H NMR spectrum of the system was closely matched with that of
1·NTf2 (Fig. S43 in Supporting information). Moreover, the
19F NMR spectra of
TBANTf2@1·OTf (
Fig. 4d and Fig. S44 in Supporting information) showed that the signals of the bound OTf
- anions on the cage diminish gradually, and the introduced NTf
2- anions take their place. These results suggest that NTf
2- anions are able to fully replace the bound OTf
- anions and also verify that NTf
2- anion has a stronger binding ability with MOC
1.
Since the NTf
2- anion is larger than the OTf
- anion, which may impact its ability to bind to the cage [
44], we opted to conduct titration experiments using the smaller BF
4- anion to validate this hypothesis. First, the BF
4- anions were titrated into the
1·NTf2 solution (marked as
TBABF4@1·NTf2), and the
1H and
19F NMR spectra of this process were recorded. As shown in Fig. S43, after adding 16 equiv. of BF
4- anions, the
1H NMR spectra of
TBABF4@1·NTf2 remained almost unchanged. Additionally, the
19F NMR spectra of
TBABF4@1·NTf2 (Fig. S45 in Supporting information) clearly revealed that the encapsulated NTf
2- anions were not displaced by BF
4- anions, and there was no obvious signal of encapsulated BF
4- anions. This proves that the binding ability of the cage to NTf
2- anion is indeed stronger than that of BF
4- anion. Next, the same titration procedure and characterization was used for the
TBABF4@1·OTf system (Figs. S47 and S48 in Supporting information). As BF
4- anions were gradually added, the peak in the imine hydrogen region of
TBABF4@1·OTf showed a slight change. This may be caused by the replacement of some OTf
- anions, which was confirmed by the
19F NMR spectra of the system. The originally encapsulated OTf
- anions in
1·OTf were partially replaced. When 16 equiv. of BF
4- anion were added, this replacement still did not reach equilibrium. These results reveal that under the same concentration, the binding ability of OTf
- anion to the cage is still stronger than that of BF
4- anion. The above titration experiments generally reveal that NTf
2- anion with larger size and charge distribution area may have stronger host-guest interactions with the cage.
The isothermal titration calorimetry (ITC) technology is often used to study the association constants and thermodynamic parameters involved in forming host-guest complexes, which is important for understanding the binding process between host and guest [
47–
50]. Upon the addition of TBAOTf, TBANTf
2, or TBABF
4 (20 × 10
-3 mol/L) to a CH
3CN solution of
1·NTf2 or
1·OTf (1.0 × 10
-3 mol/L), the association constants and thermodynamic parameters of these processes could be obtained and showed in Figs. S51 and S52 (Supporting information) and
Table 1. Comparing and analyzing the experimental results, we observed that the association constant (
K) value of the
TBANTf2@1·OTf system was the highest at 1.314 × 10
3 L/mol, while the corresponding
TBAOTf@1·NTf2 system did not exhibit significant heat exchange. This finding indicates that
1·OTf can convert to
1·NTf2 during the titration process with TBANTf
2, whereas
1·NTf2 remains unchanged during the titration process with TBAOTf. This result is highly consistent with the NMR titration data. It is worth noting that in the
TBANTf2@1·OTf system, the △
H and
T△
S values are 0.290 and 0.486 kcal/mol, respectively, indicating that the complexation process is driven by entropy. Additionally, the
K value of
TBABF4@1·NTf2 system is 0.254 × 10
3 L/mol, which suggests that there could be some anion exchange processes occurring in the system. However, due to the entropy-driven nature of the above processes, they exhibit slight variations from the NMR results.
In our previous work [
40], we have already demonstrated the selective encapsulation properties of the MOC
2 for pyrene. Hence, we are interested in investigating whether the binding ability of MOC
1 to pyrene is altered following vertex modification. To broaden the scope of our experiments, pyrene (Py) and its analogs 4,5,9,10-tetrahydropyrene (4H-Py) and hexadecahydropyrene (16H-Py) were selected as guests for binding test. In two standard experiments,
1·NTf2 and
1·OTf were dissolved in CH
3CN (0.4 mL, 0.001 mmol), respectively, and an excess (5 equiv.) of pyrene was added to each solution. The reaction systems were equilibrated at 70 ℃ for 3 h. The reaction products were analyzed by NMR spectra (
1H NMR,
19F NMR, COSY, NOESY, and DOSY, see Supporting information for details). As shown in
Fig. 5a, most signals of
1·NTf2 remained unchanged, and a new set of characteristic peaks corresponding to
Py@1·NTf2 emerged, which was also confirmed by ESI-TOF mass spectrum (Fig. S56 in Supporting information). It was worth noting that the
19F NMR spectra (
Fig. 5b) of the NTf
2- anions encapsulated in the cage did not change significantly before and after encapsulation of pyrene. The results indicate that the NTf
2- anions have a stronger binding affinity with the cage than pyrene molecules. In comparison, the
1H NMR spectra of
Py@1·OTf (
Fig. 5c) exhibited noticeable shifts in either the high-field or low-field regions upon pyrene encapsulation compared to
1·OTf. It was observed that the two sets of signals for imine atoms of
1·OTf merged into single group. Upon the release of the encapsulated OTf
- anions (
Fig. 5d), the signals for F also merged into a single set (Fig. S66 in Supporting information), indicating complete replacement of the encapsulated OTf
- anions. The ESI-TOF mass spectrum results also supported that the cage cavity was completely occupied by pyrene with the molar ratio of 1:1 for the host-guest complex (Fig. S70 in Supporting information). When pyrene is replaced by 4H-Py and 16H-Py, similar results can be obtained (see Supporting information for details). Overall, the host-guest chemistry experiments involving pyrene and its analogs further confirmed the stronger binding interaction between NTf
2- anion and the cage compared to OTf
- anion.
Since these polycyclic hydrocarbon guests can completely displace OTf
- anions from the cage, it is beneficial to use ITC technology to study the thermodynamic parameters and association constants in these processes.
1·OTf and
2·OTf were selected as host molecules for subsequent testing. The ITC data were displayed in Figs. S92-S94 (Supporting information) and
Table 2. All guest exchange processes are exothermic processes, indicating that they are driven by enthalpy. From the data listed in
Table 2, Py and its analogs form complexes with
1·OTf, and their association constants follow the order:
K(
Py@1·OTf) >
K(
4H-Py@1·OTf) >
K(
16H-Py@1·OTf). The same trend is observed for the complexes with
2·OTf. Additionally, the time-dependent
1H NMR spectra of the binding process between
1·OTf or
2·OTf and Py or its analogs were recorded (Figs. S95-S100 in Supporting information). The results indicate that the system reaches equilibrium at 5 min after adding pyrene to the
1·OTf solution. In comparison, the corresponding equilibrium times for the 4H-Py and 16H-Py systems are 1 h and > 2 h, respectively. The host-guest system corresponding to
2·OTf shows a kinetic encapsulation process similar to that of
1·OTf. These can be attributed to the fact that Py has a larger
π electron delocalization structure, which can form stronger C–H⋅⋅⋅
π interactions with the cage skeleton [
40]. However, due to partial hydrogenation, the aromaticity of 4H-Py is reduced, resulting in a decrease in the number of binding sites and strength. Fully hydrogenated 16H-Py lacks a conjugated system, has a more flexible structure and weaker binding ability. It is worth noting that there are significant differences in the binding constants for the formation of host-guest complexes with the same guest molecule, as seen in
1·OTf and
2·OTf. Taking Py as an example, its association constant with
1·OTf is 0.921 × 10
3 L/mol, and with
2·OTf it is 1.11 × 10
3 L/mol, the latter being greater than the former. The same situation occurs when 4H-Py and 16H-Py are guests. This may be due to the electron-withdrawing nature of the fluorine atoms introduced by the vertex modification, which changes the overall charge distribution of MOC
1 skeleton, weakening its interaction with Py and its analogs. To verify this speculation, electrostatic potential (ESP) analysis based on density functional theory (DFT) calculations was performed to probe the effect of F substitution on the charge distribution of the skeletons of MOC
1 and
2. As illustrated in
Fig. 6, the positively charged areas at the face and window positions of MOC
1 are more extensive and intense than those of MOC
2, indicating that the F substitution induced a decrease in the electron density of the cage skeleton. The above changes lead to a weakening of the C–H⋅⋅⋅
π interaction between the guest and the MOC
1 skeleton [
40], which is manifested as a binding constant smaller than that of the corresponding MOC
2.
In summary, we demonstrate a feasible vertex modification approach for tuning the host-guest chemistry of metal-organic cages. The vertex modification of MOC 1 with fluorine atoms significantly alters its host-guest chemistry compared to the non-fluorinated analog MOC 2. Structural and spectroscopic analyses confirmed that MOC 1 retains a hexahedral skeleton but exhibits enhanced anion selectivity, favoring larger NTf2- over OTf- or BF4-. Furthermore, pyrene displaced OTf- but not NTf2-, reflecting the hierarchical binding strengths of anions. The reduced association constants of 1·OTf with pyrene and its analogs, compared to 2·OTf, highlight the electronic effects of fluorine substitution on host-guest affinity. This work demonstrates that strategic vertex modification in MOCs enables fine-tuning of anion selectivity and guest-binding properties. This approach offers valuable insights for creating new functional materials based on metal-organic cages for applications such as molecular recognition, sensing, and catalysis in the future.
PDF
Full
77
44
关闭全屏
No. 86 Xueyuan South Road, Haidian District, Beijing
010-62662699(Operation and Technical Support)
qkjq@cast.org.cn
