In order to further confirm the structural characteristics of the macrocycles, single crystals of NP2a were obtained as yellow bulk crystals by slowly volatilizing in chloroform solution under ambient conditions, and single crystal X-ray diffraction analysis was performed. As is shown in Figure 2A, NP2a has a near-planar structure, in which the long axial length of the macrocycle is 7.491 Å while the short axial length is 5.987 Å. Taking the phenanthrene (Phen) units as the base plane, the two naphthalimide (NMI) building blocks are slightly twisted in opposite directions with a dihedral angle of 20.4°(Figure 2B). In addition, there are two parallel NP2a molecules in one unit cell and the long-range stacking of NP2a is layered with a p-π spacing of 3.744 Å (Figures 2C,D).

Many methods had also been tried to grow the single crystals of NP3a. Unfortunately, due to the twisted topology of the macrocycle, single crystals suitable for X-ray diffraction analysis were not obtained and the structure of NP3a was optimized by DFT calculations at B3LYP/6–31G (d,p) level of theory. As shown in Supplementary Figure S2, the macrocycle NP3a shows an extremely twisted geometry, and the dihedral angles between the phenanthrene (Phen) units and the naphthalimide (NMI) units vary between 50 and 55°. It can be inferred that the near-planar macrocycle NP2a constructed from building blocks with perfect bonding angles has a completely different topology from the macrocycle NP3a constructed with imperfect cyclization angles, which may be the reason for different physicochemical properties and assembly behaviors of the two macrocycles. When it comes to similar macrocycles NP2b and NP3b, we can find that they only have changes in substituents compared with NP2a and NP3a, which will not drastically affect the structure of their conjugated skeletons. The specific molecular geometries are also calculated by DFT calculations, where NP2b adopts a near-planar structure consistent with NP2a, and NP3b shows twisted ring structure similar to NP3a (Supplementary Figure S2).

To further understand the spatial geometry and electronic properties of these conjugated macrocycles, density functional theory (DFT) calculations based on first-principles were conducted. To simplify the calculation, all long alkyl groups in these macrocycles were replaced by methyl groups. As shown in Figure 3, the distribution of electron cloud in the alkyl-substituted molecules NP2b/NP3b is basically the same as that of the aryl-substituted macrocycles NP2a/NP3a. Both NP2a and NP2b show a planar structure, where the LUMOs are mainly distributed on two naphthalimide (NMI) building blocks while the HOMOs show main weight on two phenanthrene (Phen) units. On the other hand, the electron cloud distributions of NP3a and NP3b show great imbalance. The LUMOs are mainly localized on two of the naphthalimide (NMI) units while the other naphthalimide unit has almost no delocalization. Similarly, the HOMOs are distributed on two of the phenanthrene (Phen) units while the other phenanthrene (Phen) unit shows no HOMO electron cloud. These may be due to the overly distorted geometry of the macrocycles NP3a and NP3b, which greatly affects the conjugated backbone. In addition, the LUMO and HOMO levels of macrocycles can also be calculated (Shetty et al., 1996), in which the band gaps of the trimers NP3a/NP3b are all larger than the corresponding dimer NP2a/NP2b, which can also be attributed to the twisted molecular structure.

The newly synthesized macrocycles are all yellow solids with good solubility, which can be dissolved in most organic solvents, such as toluene, chloroform, dichloromethane and tetrahydrofuran. The absorption spectra and fluorescence spectra of NP2a, NP3a, NP2b and NP3b were measured in toluene (ca. 10⁻⁵ M). As shown in Figure 4A, there are two main absorption peaks in the absorption spectrum of each macrocycle, where the peak in the high-energy region corresponds to the p-π* transition of electrons in the conjugated backbone of macrocycle while the absorption peak in the low-energy region corresponds to the charge transfer from the electron-donor phenanthrene (Phen) units to the electron-acceptor naphthalimide (NMI) units. In addition, the molar extinction coefficients and optical band gaps of four macrocycles were calculated according to the absorption spectra, which are shown in Table 1.

Comparing the conjugated macrocycles NP2a and NP3a, it is found that NP3a has a larger molar extinction coefficient than NP2a, which indicates that the conjugated system of NP3a has been expanded. However, the optical band gap of NP3a decreases compared to that of NP2a, which may be attributed to that the distorted structure of NP3a affects the continuity of the conjugated system, weakening the conjugation to a certain extent. The same phenomenon is also observed in alkyl-substituted macrocycles NP2b and NP3b. In addition, it is interesting to find that the aryl-substituted macrocycle NP2a have a larger molar extinction coefficient than the alkyl-substituted macrocycle NP2b although the trend of the absorption spectrum has no obvious change. This indicates that though the conjugated backbones of the two macrocycles have not changed significantly, the different substituents may affect the behavior of molecules in solution, resulting in different molar extinction coefficients. The above phenomenon also occurs in NP3a and NP3b.

Different from the absorption spectra, the fluorescence spectra of the four conjugated macrocycles in toluene solution are basically the same and the maximum emission wavelengths are all around 516 nm, which can be deduced that the influence of the substituents as well as the expansion of the conjugated system on the fluorescence emission is negligible for the four macrocycles (Figure 4B ). The absolute fluorescence quantum yields of NP2a, NP3a, NP2b and NP3b in solution were determined to be 30.8%, 28.85%, 25.9% and 25.57% respectively, by using the integrating sphere technique. The absolute fluorescence quantum yields of NP2a, NP3a, NP2b and NP3b in solid were also estimated as 5.17%, 9.79%, 1.40% and 2.62% relatively. The difference may be attributed to the solid phase stacking of the macrocycles.

In order to further investigate the intramolecular charge transfer (ICT) of the macrocycles, the solvatochromic effects on the absorption and PL features were investigated. It was found that both four kinds of п-conjugated macrocycles had similar solvatochromic effects (Supplementary Figures S3, S4). For example, although the absorption spectra of NP2a in different solvents showed negligible change, its fluorescence spectra showed remarkable solvation effect, indicating that there was significant donor-acceptor electron transfer in the macrocycle. And the more obvious solvation effect in fluorescence spectra than absorption spectrum may be due to that the excited electrons are more easily polarized (Zhu et al., 2018). In addition, cyclic voltammetry (CV) measurements of macrocycles NP2a, NP3a, NP2b and NP3b were also carried out in a solution of tetrabutylammonium hexafluorophosphate (n-Bu₄NPF₆) in anhydrous DCM with a scan rate of 50 mV s⁻¹. As shown in Supplementary Figure S5, all the four macrocycles display similiar irreversible redox waves. From the onset potential of the first reductive wave, the LUMO energy levels of NP2a, NP3a, NP2b and NP3b are estimated as −3.02, -3.03, -3.05 and -3.02 eV, respectively.

It is found that the aromatic chemical shifts of NP2b and NP3b were dependent on concentration when ¹H NMR spectroscopy was conducted. What’s more, the different absorption phenomena of macrocycles with different substituents also implied that there may be a special self-assembly phenomenon in solution. As is shown in Figure 5A, the chemical shift of aromatic proton in the lowest field of NP2b varied from δ = 9.14 to 8.42 ppm as the concentration changed from 1.13 to 21 mM. Similar phenomena was also been observed in NP3b, where the proton varied from 8.88 to 8.73 ppm in the same concentration range (Figure 5B). All of these suggest that the alkyl-substituted macrocycles self assemble with π–π stacking and the upfield shifts of the aromatic protons can be attributed to the influence of ring-current magnetic anisotropy of the adjacent macrocycle (Mao et al., 2020); (Shetty et al., 1996). In contract, the chemical shifts did not change significantly even when the concentration of aryl-substituted NP2a/NP3a was increased from 1.13 to 21 mM (Supplementary Figure S3), implying negligible self-aggregation, which may be due to the fact that the rigid aryl substituents greatly restrict the p-π stacking of macrocycles in solution.

Using the monomer-dimer model (Zhao and Moore, 2003) to simulate the chemical shift as a function of concentration, the self-association constants of NP2b and NP3b can be obtained. This model assumes that monomer-dimer equilibrium is the predominant process and higher order aggregations are negligible, which is sufficient for self-assembled systems in moderately polar chlorinated hydrocarbon solvents such as chloroform. The association constants for the self-aggregation of NP2b and NP3b were fitted by a computational method developed by Horman and Dreux (Horman and Dreux, 1984). As shown in Figure 5, the self-aggregating association constants of NP2b and NP3b are 65.2 and 6.18 L mol⁻¹, respectively (Figures 5C,D). In addition, if macrocycles that can self-assemble in solution are applied to thin-film field-effect transistors, better performance may be obtained (Luo et al., 2010) (He et al., 2013), and these further studies are also underway in our laboratory.

There is an inner cavity with a defined size in the macrocycle, which is beneficial for it to act as the host and combine with some guest molecules to achieve supramolecular binding. Here, C₇₀ was selected as the typical guest molecule, and was added dropwise to the deuterated chloroform solutions of NP2a, NP3a, NP2b and NP3b, respectively. The mixture was tested by ¹H NMR at room temperature and the supramolecular binding behavior was characterized by chemical shift.

As shown in Figure 6A, when C₇₀ was dropped into the solution of macrocycle NP2a, the two groups of aromatic protons in the macrocycle gradually separated from the original overlapping state. In addition to the obvious separation of protons, the aromatic protons in NP2a also showed a slight upfield shift trend with the addition of C₇₀. All of these observations indicated that there was an obvious interaction between the macrocycle NP2a and C₇₀, which significantly changed the chemical environment of the protons in the original solution of NP2a. Similar dimer macrocycle NP2b also showed changes, in which the aromatic proton shifted slightly to the downfield from δ = 8.96–9.01 ppm as the addition of C₇₀ increased from 0 eq to 4eq (Figure 6C), demonstrating that there is also a weak supramolecular interaction between the host NP2b and guest C₇₀. In contract, the trimer macrocycles NP3a and NP3b exhibited more pronounced chemical shift changes than dimer macrocycles NP2a and NP2b (Figures 6B,D). The chemical shift of the protons of NP3a and NP3b varied from 8.95 to 8.90 ppm and from 8.81 to 8.75 ppm, respectively, as the addition of C₇₀ increased from 0 eq to 4eq.

In order to quantitatively study the supramolecular assembly of macrocycles with fullerenes, C₇₀ was added to the toluene solution of macrocycles NP2a, NP3a, NP2b and NP3b (ca. 10⁻⁵ M) respectively and the change of the fluorescence emission spectrum of the mixture was monitored. As shown in Figure 7, with the continuous addition of C₇₀, the maximum emission peaks of the macrocycles NP2a and NP3a were all quenched to varying degrees, indicating that supramolecular complexes had been produced in solution. Similar fluorescence quenching was also observed in toluene solutions of NP2b and NP3b with the addition of C₇₀ (Supplementary Figure S4). By a mole ratio plot based on fluorescence emission spectrum data (Supplementary Figure S7), a 1:1 stoichiometry was confirmed for NP2a/NP3a/NP2b/NP3b and C₇₀ (Thordarson, 2011) (Brynn Hibbert and Thordarson, 2016) (Li et al., 2014). According to the non-linear curve-fitting method (Yang et al., 2020), the binding constants of NP2a, NP2b, NP3a and NP3b with C₇₀ were calculated to be 5.89×10⁴ M⁻¹, 2.68×10⁴ M⁻¹, 1.56×10⁵ M⁻¹ and 3.14×10⁴ M⁻¹respectively, indicating that trimers NP3a and NP3b have stronger binding to C₇₀ than dimers NP2a and NP2b, which is consistent with the results of the ¹H NMR titration test. The stronger supramolecular assembly with C₇₀ may be related to the larger internal cavity and non-planar topology of macrocycles NP3a and NP3b.

Considering that the macrocycles are consisting with electron-deficient naphthalimide units, the supramolecular assembly between NP2a/NP3a/NP2b/NP3b and the electron-rich guest pyrene was also investigated. The concentration of the macrocycles was kept constant, and the guest pyrene was added at room temperature according to the equivalent. The variation of characteristic peaks was monitored by NMR (Supplementary Figure S9, S10). The results show that only the planar macrocycles NP2a and NP2b are obviously encapsulated with pyrene, while the ¹H NMR spectra of NP3a and NP3b changed little, indicating that they could not bind with pyrene. This may be due to the distorted geometry of NP3a/NP3b which is not suitable for binding with planar molecules. By a mole ratio plot based on chemical shift data, a 1:1 stoichiometry was confirmed for NP2a/NP2b and pyrene (Zeng et al., 2022). Through non-linear fitting, the binding constants of NP2a/NP2b with pyrene are 327.43 and 253.33 M⁻¹ respectively.