Order from Chaos: Self-Assembly of Nanoprism from a Mixture of Tetratopic Terpyridine-Porphyrin Conformers

Summary of main observation and conclusion

Porphyrins have been widely used in the self-assembly of metallo-supramolecules. In this study, we introduced 2,2’:6,2”-terpyridine (tpy) into a porphyrin core to synthesize a tetratopic building block with multiple conformers. During the self-assembly with Zn(II), such a mixture of conformers was able to form a discrete nanoprism with all building blocks in one conformation. Detailed characterizations, including NMR, ESI-MS and traveling-wave ion mobility-mass spectrometry (TWIM-MS), all supported the formation of the desired assemblies. AFM and TEM further confirmed the dimensions of assembled nanoprisms. Moreover, the photophysical properties of the ligands and complexes were noticeably different depending upon size and metal ion center.

Background and Originality Content

Porphyrins are a vibrant class of organic dyes found widely in nature, ranging from the well-studied chlorophyll and heme found in the cells of plants and red blooded animals to the less known chlorocruoroporphyrin found in annelids.^[1] Porphyrins in natural systems are as diverse in structure as they are in function. Heme^[2] and chlorocruoroporphyrin^[3] function as the core of oxygen transport proteins, which is responsible for transporting oxygen to vital systems; while chlorophyll provides necessary electron transport functions required for photosynthesis. As a result of millions of years of successive improvement, nature has perfected the function of porphyrins. In the field of synthetic chemistry, substitution and modification have been used to elicit specific functions of porphyrins. Free base porphyrins have been extensively studied in the biomedical fields, e.g., clinical phototherapy and cancer imaging.^[4] Metallo-porphyrins have been applied in sensors,^[5] molecular electronics,^[6] photovoltaic devices,^[7] iono-phores^[8] and catalysts.^[9]

Much work has been done to increase the size, complexity and function of porphyrins with the goal of expanding the scope of porphyrin chemistry, using conventional covalent bonding and supramolecular self-assembly. For example, porphyrin tweezers^[10] were constructed through the coordination with metal ions to control the capture and release of guest molecules. Supramolecular porphyrin boxes^[11] and cages^[12] with unique photodynamic properties could be applied to the fields of energy harvesting and cancer phototherapy. Porphyrin macrocycles^[13] and prisms^[14] were able to encapsulate large guest molecules and to achieve enrichment of specific fullerenes with high selectivity.^[14b]

In coordination-driven self-assembly, most of the building blocks possess rigid geometry with single dominant conformation, which is able to prevent mismatching of the ligands, such as unwanted polymerization during the self-assembly.^[15] It is rare to assemble discrete metallo-supramolecules using building blocks with multiple conformations. Herein, we report the construction of a porphyrin nanoprism through self-assembly of Zn(II) with 5,10,15,20-tetrakis(3-([2,2’:6’,2”-terpyridin]-4’-yl)phenyl)-7H,8H-porphyrin, which contains multiple conformational isomers. During the self-assembly, all the building blocks were converted into one conformation through coordination to achieve order from chaos. Detailed structural characterization confirmed that the conformations of building blocks were fixed within the metallo-supramolecular architecture. Further study showed that the photophysical properties of the ligands and complexes were noticeably different depending upon size and metal ion center.

Results and Discussion

Scheme 1a depicts the synthesis of a tetratopic tpy-porphyrin ligand (L1) through established methods.^[16] Particularly, the porphyrin core was synthesized using the Adler-Longo method.^[17] The metalated ligand (L2) was obtained by treating L1 with 5 equivalents of zinc acetate in a 19 : 1 solution of chloroform and methanol followed by chelation with EDTA to remove the excess amount of Zn(II) attached to the tpy groups. It is worth noting that a longer version of tetratopic tpy-porphyrin was reported but with different synthetic strategy.^[18] Our approach is a metal free synthesis, which avoids the multi-armed defects during Suzuki coupling and improves the separation.

Scheme 1.

(a) Synthesis of tetratopic tpy-porphyrin ligands L1 and L2. (B) Cartoons for the four conformers of L1 (L1-A to L1-D) with different orientations of tpy

NMR results of L1 (Figure 1a, Figures S2–S5) provide evidence supporting the existence of isomers (Scheme 1b), which come from the steric repulsion of H^A-H^B and H^A-H^D during the rotation of tpy groups along phenyl-porphyrin C—C bond. Figure 2a displays a messy spectrum, especially in the region of δ 8.5–8.7, where the peaks are assigned to H^G and H^J on the tpy group. Such NMR shifts indicate the tpy groups in the tetratopic ligand are settled in various chemical environments, which are further supported by at least three distinct sets of H^I-H^J correlation signals in 2D-COSY spectra (Figure S5). Furthermore, we utilized traveling wave ion mobility-mass spectrometry (TWIM-MS)^[19] to address the conformers in L1. Figure 2a shows the drift time distributions obtained from ion mobility separation of ions at m/z 771, which corresponds to the [L1+2H⁺]²⁺ species according to the isotope patterns (Figure 2b). Four distinct signal peaks at different drift times (i.e., 2.70, 2.84, 3.05, and 3.32 ms, respectively) are observed in the spectrum, and correspond to four conformers of L1 with different collision cross-section areas. Similarly, multiple sets of tpy proton resonance peaks (Figure 1c, Figures S6–S9) in ¹H NMR and multi-distributed signal bands in TWIM-MS (Figure S1) were also observed in L2, indicating that L2 is also a mixture of conformers. The energy barrier for the rotation of one arm was studied by conformational search (Figure S31, detailed method was summarized in SI), which was calculated as 72.44 kJ-mol⁻¹. It is higher than the rotational energy of C—C in freely rotating n-butane (19 kJ-mol⁻¹)^[20] but lower than that of conformationally stabilized 1,1’-bi-2-naphthol (167 kJ-mol⁻¹).^[21] Compared to the reported longer version of tetratopic tpy-porphyrin ligand, our design introduced more restriction to the rotation of tpy with a short arm, and constructed more compact structures with higher stability.^[18]

Figure 1.

¹H NMR spectra of (aromic region, 400 MHz, 300 K) (a) L1 (in CDCl₃), (b) S1 (ca. 1 mg/mL in CD₃CN), (c) L2 (in CDCl₃), and (d) S2 (ca. 1 mg/mL in CD₃CN) and 2D-DOSY spectra of (500 MHz, CDCl₃, 300 K) of (e) S1 and (f) S2 (aromic region). * Representing CHCl₃ in the solvent.

Figure 2.

(a) TWIM-MS spectrum of ions at m/z 771 for L1. Ion mobility separation gave signals at 2.70, 2.84, 3.05, and 3.32 ms, corresponding to four conformers of [L1+2H⁺]²⁺, respectively. (b) Theoretical and experimental isotope patterns of the four separated peaks.

Because of the dynamic nature of those four conformers, it is almost impossible to separate the conformers using regular column chromatography. Among the four possible conformers (L1-A to L1-D, Scheme 1b), L1-A could be the only candidate for constructing a discrete structure via coordination-driven self-assembly because of its introverted conformation. As a result, if we directly assemble the mixture of ligands with the stoichiometric amount of Zn(II), three circumstances could occur: (i) formation of polymeric assemblies; (ii) coexistence of discrete supramolecular architecture together with polymers; and (iii) generation of a pure discrete structure via conformational transformation from L1-B/C/D to L1-A during the assembly process.

After performing the assembly in CHCl₃/MeOH (V/V = 1/3) at 50 °C for 5 h, the product was precipitated by adding excessive NH₄PF₆ in MeOH solution, followed by several washings with water. Then the assemblies of S1 and S2 were obtained in high yields from L1 and L2, respectively (Figures 3a and 3b), for further characterization. ESI-MS spectrum of S1 (Figure 4a) displays one dominant set of peaks with continuous charge states from 6+ to 10+. The deconvoluted molecular mass of the assembly is 6749.8 Da, which agrees well with the molecular composition of three ligands, six zinc ions, and 12 PF₆⁻ as counterions, viz., [(C₁₀₄H₆₆N₁₆)₃Zn₆-(PF₆)₁₂]. The experimental isotope patterns are consistent with the theoretical ones based upon the chemical composition of S1 (Figure 4a inset, Figure S24). We therefore proposed a supramolecular nanoprism with the simulated structure shown in Figures 3c and 3d. The possi- bility of forming the Zn-porphyrin (Zn-Por) complexes during the self-assembly of S1 can be excluded by carefully controlling the amount of Zn(II). The undesired complexation does not occur because of the preference of forming the Zn-tpy complex rather than Zn-Por complex, even under the heating process, which was confirmed during our synthesis of L2.

Figure 3.

Self-assembly of (a) supramolecular nanoprism S1 and (b) S2; representative energy-minimized structures of S1 (c, d) and S2 (e, f) by molecular modeling.

Figure 4.

ESI-MS of (a) S1 and (c) S2. Insets are the theoretical and experimental isotope pattern of the species with 7+ charge state; TWIM-MS plots (m/z vs. drift time) of (b) S1 and (d) S2.

ESI-MS of S2 (Figure 4b) gave similar results and the corresponding isotope patterns are shown in Figure 4b inset and Figure S25, suggesting that a similar nanoprism was assembled (Figures 3e and 3f). In addition, a single band of signal at each charge state with narrow distribution is displayed in the TWIM-MS spectra of S1 (Figure 4b) and S2 (Figure 4d), which excludes the existence of isomers after the self-assembly. These results indicate that all other conformers of L1/L2 were converted to the introverted structure during the self-assembly process to form a discrete nanoprism structure (S1/S2).

Then multi-dimensional NMR spectroscopy was utilized to further identify the structure and purity of S1 and S2 (Figure 1, Figures S10–S23). After the self-assembly, the sharp and well-split peaks of supramolecules S1 and S2 instead of broad signals of L1 and L2 suggest that a discrete structure rather than polymeric assemblies is obtained. 2D-DOSY spectra of S1 and S2 (Figures 1e and 1f) display single narrowly distributed signal bands with diffusion coefficients (D) around 1.2 × 10⁻⁹ m²·s⁻¹ (S1) and 1.0 × 10⁻⁹ m² s⁻¹ (S2), respectively, further confirming the isomeric purity of the assemblies. More detailed analysis of the ¹H NMR spectra of L2 and S2, for instance, shows characteristic downfield shift of H^f (δ ~0.05) and obvious upfield shift of H^j (δ ~1.0) after coordination, due to the electron deficiency of H^f and the electron shielding effect of H^j by the zinc ions,^[22] respectively. In addition, the shifts of the protons on the phenyl spacer after coordination were observed.

The downfield shifts of H^a,c,e (δ 0.2–0.4) were caused by the electron withdrawing effect from the coordination sites, while the abnormal electron shielding on H^d (δ ~0.1 towards upfield) is probably because the protons are located in the lateral faces of the nanoprism (Figure 3e). Notably, after coordination, the resonance signal of H^a was split into two sets after self-assembling into S2, indicating that H^a settled in two different environments, i.e., the base surface (H^a1) and lateral surface (H^a2) of the prism. The same characteristic shifts and the splitting of H^A after assembly are also observed in the spectrum of S1.

A controlled self-assembly of L1/Zn(II) at lower temperature (0 °C) was conducted. The ¹H NMR spectrum (Figure S32a) of the complexes assembled at lower temperature displays a great number of undesired peaks in addition to the peaks of S1, which proves the formation of polymeric assemblies. The mixture was self-corrected to form pure S1 after 3 h of heating at 50 °C (Figure S32b), owing to the dynamic transition among the L1-A to L1-D at higher temperature.

After removing the tpy-coordinated Zn(II) ions in S1 and S2 by mixing the supramolecular solution with EDTA under room temperature, the ¹H NMR spectra of the recovered ligands after chromatographic purification show exactly the same spectra with those of L1/L2 before the assemblies. This further indicates that the transition potentials among the conformers are relatively low and dynamic, and thus drive the conformers into adopting the introverted conformation for forming the thermodynamically controlled nano-prism structure during the assembly.

Atomic Force Microscopy (AFM) and Transmission Electron Microscopy (TEM) were used to image individual supramolecular nanoprism. Individual dots were observed in both TEM and AFM images of S1 (Figure 5) with a uniform molecular size. In the AFM images (Figures 5a–5c), the supramolecules are uniformly distributed. Using the section analysis on AFM, the height of the supramolecule is ca. 1.9 nm which shows a good correlation with the energy minimized model (Figure 3c). The AFM and TEM images of S2 show similar results, which are summarized in Figure S26.

Figure 5.

(a) AFM image of S1, (b) 3D AFM image of S1, (c) Cross-section of one randomly selected particle in (a), (d) TEM images of S1, inset: zoomed-in TEM images of the selected area.

Concentration-dependence of forming S1 and S2 was carefully studied. ¹H NMR spectra of S1 in various concentrations (Figure S27) show obvious changes by gradually decreasing the concentration. For instance, the peaks at σ 7.70, 9.19 and 9.30 shown at high concentration (5 mg/mL) gradually shrunk by lowering the concentration. Only one dominant set of peaks is displayed in the diluted (0.62 mg/mL) sample. Such a concentration-dependent variation was also observed in ESI-MS spectra (Figure S29), which helps us further determine all assembled structures. At a higher concentration (2.5 mg/mL), besides the dominant set of peaks assigned to S1, another two minor sets of peaks are also clearly observed, which are ascribed to tetramer (M4, L1₄Zn₈) and pentamer (M5, L1₅Zn₁₀), respectively. These two sets of signals gradually shrink and disappear by diluting the sample solution. The experiment shows that S1 is the major structure in diluted solution.

However, no significant change was observed in the concentration-dependent experiments of S2, which showed one dominant set of signals in ¹H NMR spectra (Figure S28) and in ESI-MS spectra (Figure S30) under all concentrations. The significant difference in concentration-dependent behavior of S1 and S2 might come from the different flexibility of Zn free porphyrin and Zn-Por backbones. After coordination with Zn, the porphyrin ring backbone becomes more rigid and restricts any adjustment required to form the larger membered prisms.

Absorption and emission spectra of the ligands and supra-molecules were collected in CHCl₃/CH₃CN (V/V = 1/9) mixed solvent for all ligands and supramolecules (Figure 6). The absorption spectra show a strong absorption band at around 420 nm, corresponding to the Soret bands of the porphyrinyl group. The indicative four Q bands are displayed in the UV-Vis of L1 and S1, respectively, around 500–700 nm.^[23] In the case of L2 and S2, when Zn(II) is inserted into the porphyrin ring, the UV-Vis spectra exhibit two peaks instead of four in the Q regions, due to the modification in the symmetry of the substituted porphyrin ring from D_2h of L1/S1 to the D_4h of L2/S2.^[23] Two emission bands at 650 nm and 720 nm for L1/S1 and 600 nm and 650 nm for L2/S2 are observed in the emission spectra (Figure 6c). Obvious weaker emission spectra are shown in L2/S2 compared with those of L1/S1, due to the additional zinc atom in the center of the emissive porphyrinyl chromophore.

Figure 6.

UV-Vis absorption spectra for (a) L1 and S1, (b) L2 and S2 (c = 10.0 μmol L⁻¹, solvent: 1 : 9 CHCI3/CH3CN); (c) émission spectra for L1, L2, S1 and S2 (λ_ex = 420 nm, c = 10.0 μmol L⁻¹, solvent: 1 : 9 CHCI₃/CH₃CN).

Conclusions

Two new porphyrin-based tetratopic tpy ligands, L1 and L2, were synthesized and proved to have multiple conformers under ambient temperature. Discrete supramolecular nanoprisms, S1 and S2, instead of a mixture of polymers, were obtained through coordination-driven self-assembly from the mixture of conformers. Due to the dynamic transformation among these conformers, nanoprisms were constructed as the energy favorable structures by fixing all the ligands in one conformation. Therefore, this study provides another example of “order-out-of-chaos” through self-assembly.

Experimental

L1: P3 (2.19 g, 6.5 mmol) and propionic acid (20 mL) were added into a round bottom flask. The mixture was heated under 60 °C, pyrrole (0.52 g, 7.8 mmol) was added to the mixture, then the mixture was heated at 145 °C for 3.5 h and cooled to room temperature. A 4 : 1 mixture of acetone and aqueous ammonia solution was added, and the flask was kept in a freezer overnight. The resultant precipitate was collected and washed with cold methanol 3 times. After being dried, the solid was further purified by chromatography on SiO₂ column with CHCl₃ as the eluent. 135 mg purple solid was obtained with a yield of 6%. ¹H NMR (400 MHz, CDCl₃) δ: 9.00 (s, 16H, H^A,F), 8.85 (s, 4H, H^D), 8.69–8.54 (m, 8H, H^G,J), 8.39–8.31 (m, 8H, H^B,E), 7.95 (t, J = 7.9 Hz, 4H, H^C), 7.85 (t, J = 8.5 Hz, 8H, H^H), 7.3 (t, J = 7.2 Hz, 8H, H¹), −2.60 (s, 2H, H^K) ¹³C NMR (100 MHz, CDCl3) δ: 159.16, 156.15, 156.08, 150.17, 149.06, 142.95, 137.01, 136.71, 135.10, 133.18, 127.33, 126.76, 123.69, 121.25, 119.77, 119.31. ESI-MS (m/z): Calculated for [C₁₀₄H₆₆N₁₆+2H]²⁺: 770.2906; found 770.2773.

L2: L1 (61.7 mg, 0.040 mmol) and Zn(OAc)₂6H₂O (59.6 mg, 0.2 mmol) were dissolved in a mixture of 19 mL CHCl₃ and 1 mL methanol, and the resultant mixture was heated for 5 h at 60 °C. The solvent was evaporated, and the residue was dissolved in DMF (20 mL). EDTA (59.6 mg, 0.16 mmol) was added into the DMF solution, and the mixture was stirred at room temperature for 2 d. The mixture was suspended in water and extracted by using CHCl₃. The organic phases were combined and evaporated. The residue was further purified by washing with cold acetone. 22.5 mg purple solid was obtained with a yield of 35%. ¹H NMR (400 MHz, CDCl₃) δ: 9.05 (s, 8H, H^f), 8.95 (s, 8H, H^a), 8.85 (s, 4H, H^d), 8.69–8.52 (m, 8H, H^gJ), 8.40–8.30 (m, 8H, H^b,e), 7.95 (t, J = 7.9 Hz, 4H, H^c), 7.85 (t, J = 9.7 Hz, 8H, H^h), 7.3 (t, J = 7.2 Hz, 8H, Hⁱ). ¹³C NMR (125 MHz, CDCl3) δ: 156.15, 156.06, 156.01, 155.94, 150.32, 150.26, 149.01, 148.96, 143.79, 136.82, 136.77, 136.72, 136.68, 135.08, 132.98, 132.15, 127.17, 126.45, 123.76, 123.70, 123.64, 121.35, 121.30, 121.25, 120.53, 119.35, 119.30. ESI-MS (m/z): Calculated for [C₁₀₄H₆₄N₁₆Zn+2H]²⁺: 801.7474; found: 801.7339.

S1: L1 (3.0 mg, 1.9 μmol) was dissolved in 4 mL MeOH/CHCl₃ (3/1, V/V) solvent, and Zn(NO₃)₂-6H₂O (1.2 mg, 3.9 μmol) was added in the above solution. The mixture was heated at 50 °C for 5 h, then precipitated in 10 mL NH₄PF₆ (80 mg 0.49 mmol) solution in MeOH. The solid was collected by centrifugation, washed with fresh methanol and water. After drying, 4.0 mg purple solid was obtained with a yield of 93%. ¹H NMR (500 MHz, CD₃CN) δ: 9.18 (s, 4H, H^F), 9.11 (s, 4H, H^F), 9.00 (s, 8H, H^A), 8.87 (d, J = 7.7 Hz, 4H, H^E), 8.69 (s, 4H, H^D), 8.57 (d, J = 8.0 Hz, 4H, H^B), 8.46 (d, J = 8.0 Hz, 8H, H⁶), 8.31 (t, J = 7.7 Hz, 4H, H^C), 7.62–7.51 (m, 16H, H^J,H), 6.78 (dd, J = 7.7, 5.3 Hz, 8H, H^I), −2.75 (s, 2H, H^K). ¹³C NMR (126 MHz, CD₃CN) δ: 157.21, 149.90, 148.11, 147.93, 143.42, 141.08, 135.81, 128.81, 127.90, 127.43, 123.46, 122.85, 120.18. ESI-MS (m/z): 979.72 [M-6PF₆⁻]⁶⁺ (calcd m/z: 979.68), 818.87 [M-7PF₆⁻]⁷⁺ (calcd m/z: 818.87), 698.25 [M-8PF₆⁻]⁸⁺ (calcd m/z: 698.27), 604.55 [M-9PF₆⁻]⁹⁺ (calcd m/z: 604.5S), 529.59 [M-10PF₆⁻]¹⁰⁺ (calcd m/z: 529.62).

S2: L2 (3.1 mg, 1.9 μmol) was dissolved in 4 mL MeOH/CHCl₃ (3/1, V/V) solvent, and Zn(NO₃)₂ 6H₂O (1.1 mg, 3.8 μmol) was added in the above solution. The mixture was heated at 50 °C for 5 h, then precipitated in 10 mL NH₄PF₆ (80 mg 0.49 mmol) solution in MeOH. The solid was collected by centrifugation, washed with fresh methanol and water. After drying, 3.9 mg purple solid was obtained, with a yield of 89%. ¹H NMR (500 MHz, CD₃CN) δ: 9.03 (s, 4H, H), 9.01 (s, 4H, H^f), 8.91 (s, 8H, H^a), 8.73 (d, J = 7.6 Hz, 4H, H^e), 8.55 (s, 4H, H^d), 8.44 (d, J = 7.9 Hz, 4H, H^b), 8.37 (d, J = 8.1 Hz, 8H, H^g), 8.17 (t, J = 7.7 Hz, 4H, H^c), 7.59–7.48 (m, 16H,H^j,h), 7.59–7.48 (m, 16H, H^j,h), 6.70 (dd, J = 7.9, 5.1 Hz, 8H, Hⁱ). ¹³C NMR (125 MHz, CD3CN) δ: 156.99, 150.47, 149.89, 149.52, 147.73, 147.60, 144.32, 140.71, 135.83, 135.41, 135.15, 132.50, 132.10, 128.09, 127.05, 123.08, 122.70, 122.50, 120.14. ESI-MS (m/z): 846.08 [M-7PF₆⁻]⁷⁺ (calcd m/z: 816.12), 722.18 [M-8PF₆⁻]⁸⁺ (calcd m/z: 722.23), 626.04 [M-9PF₆⁻]⁹⁺ (calcd m/z: 625.99), 548.63 [M-10PF₆⁻]¹⁰⁺ (calcd m/z: 548.59), 485.57 [M-11PF₆⁻]¹¹⁺ (calcd m/z: 485.54).