Construction and interconversion of anion-coordination-based (‘aniono’) grids and double helicates modulated by counter-cations

‘Aniono’ double helicates and grids were constructed using PO₄^3– anions and a bis–tris(urea) ligand and interconverted by changing the counter-cation.

Abstract

Supramolecular assembly of well-defined discrete architectures has been of great interest due to the tunable properties of these structures in functional materials and bio-mimicking. While metal-coordination-driven assembly has been extensively studied, anion-coordination-driven assembly (ACDA) is just emerging for constructing complex supramolecular structures. Herein two A_2nL_2n (A = anion, L = ligand; n = 1 or 2) ‘aniono’-supramolecular assemblies, i.e. double helicates and the first anion grid, have been constructed based on the coordination between phosphate (PO₄^3–) anion and a bis–tris(urea) ligand. Moreover, the aniono-grid and double helicate motifs can be readily interconverted under ambient conditions by simply changing the counter-cation. These results redefine the power and scope of ACDA, which may represent a new approach in the assembly of well-defined architectures in parallel with the metal coordination-driven assembly of metallo-supramolecules.

Introduction

Assembly of supramolecular architectures plays an active role in the fields of functional materials and bio-mimicking.1 A broad array of discrete structures has been constructed based on metal-coordination-driven assembly, such as helicates, polygons, polyhedra, cages, and interlocked metallo-supramolecules.2 Meanwhile, such architectures can be accessed by (dynamic) covalent bonding3 or hydrogen bonding.4 During the assembling process, interconversion between two topologies in some systems, e.g., double helicates and mononuclear complexes,5 tetrahedral cages and triple helicates,6 and many others,7 has also been studied in terms of the dynamic functions of different structures.8 Among the various entities, helical and grid complexes are highly interesting due to their specific topologies and properties.9 As mimics of the helical motifs in DNA and proteins, synthetic helicates have fascinated supramolecular chemists with applications in molecular recognition, catalysis, molecular machines, optics, etc.10 With specific arrangement of metal ions within a square or rectangular grid, metallo-grids have been explored in magnetic and electrochemical materials as well as in supramolecular host–guest chemistry.11

Compared to the flourishing metal-coordination-driven assembly, anion-coordination-driven assembly (ACDA) is less developed, yet it provides an efficient approach to construct well-defined structures.12 In particular, the coordination between phosphate ions (PO₄^3–) and oligourea ligands has proven to be surprisingly stable.13 As such, we were able to construct a series of anion-based architectures with phosphate ions as the coordination centers and oligo-bis(urea) as the ligands, such as the first A₂L₃ triple helicate and both face-directed A₄L₄- and edge-directed A₄L₆-type (A denotes anion) tetrahedral cages.14 More recently, we further demonstrated the interconversion between the A₂L₃ triple helicate and A₄L₆ cage.15 These results imply that the ‘aniono’-supramolecular assemblies might become a new class of superstructures in parallel with the ‘metallo’-systems. In addition, held together by multiple hydrogen bonds, these metal-free ‘aniono’-systems are more bio-compatible, and their assembly–disassembly and isomerization may be readily modulated under mild conditions. However, compared to the well-established metal coordination, there are far fewer known anion complexes that exhibit the stability, predictable geometry, and symmetry needed to form such assemblies.16 These features render ACDA very attractive but also highly challenging.

In the pursuit of more complex anion-based systems, we move forward from the bis–bis(urea)17 to the bis–tris(urea) unit.18 The tris(urea) subunit exhibits stronger coordination affinity than the bis(urea) and forms stable AL₂ phosphate complexes with binding constants even comparable to those of some metal complexes.18a In addition, the coordination geometry of these complexes is analogous to the 2 : 1 metal complexes of terpyridine (tpy) (Scheme 1a), a versatile subunit for metal-based supramolecular assembly.19 This analogy suggests that the 1 : 2 phosphate–tris(urea) complex is a potentially viable motif for anion-directed assembly with more predictable structures. Moreover, symmetry-constrained, de novo structure-based design methods20 suggested that the p-xylylene-linked bis–tris(urea) L exhibited a conformation that complemented an A₂L₂ helical assembly (Fig. S1†). In this work, ligand L readily forms the double helicate and first anion-directed grid complex. Herein we report the crystal structures and solution assembly of these assemblies (Scheme 1b). The double helicate motif features a large, partially open cavity that is ideal for guest encapsulation and exchange, much like the grooves of DNA. Moreover, the grid and double helicate are easily interconverted by changing the counter-cation. Note that the switch between the M₂L₂ double helicates and 2 × 2 (M₄L₄) grid complexes had not been reported until recently when Lehn et al. described the first example of such an interconversion between a binuclear Cu^I double helicate and a tetranuclear Cu^II grid.21 Hence, the current work also represents the first ‘anion version’ of the double helicate/grid interconversion and broadens the spectrum of ACDA in supramolecular transformation.

Scheme 1. (a) Design of bis–bis(urea) and bis–tris(urea) ligands by mimicking oligopyridine–metal coordination; (b) triple helicate assembled by bis–bis(urea) ligands and TMA₃(PO₄);14a A₂L₂ double helicate (complexes 1a–1d) and A₄L₄ grid (complex 2) (A = anion, L = ligand) assembled by the bis–tris(urea) ligand L and PO₄^3– in the presence of different tetraalkylammonium cations.

Results and discussion

Synthesis, structures, and solution studies of the double helicate 1a and grid complex 2

Complex 1a

Treatment of ligand L with 1 equiv. of (TMA)₃PO₄ (TMA⁺ = tetramethylammonium) in acetonitrile, followed by slow vapor diffusion of diethyl ether, afforded yellow block crystals of (TMA)₅[(TMA)⊂(PO₄)₂L₂] (1a). As expected, a two-stranded structure is obtained. However, instead of the predicted C_i symmetric double mesocate geometry (Fig. S1b†), complex 1a exhibits less symmetric helical A₂L₂ assemblies, which might originate from the non-centrosymmetric tetrahedral TMA⁺ guest (Fig. 1). The two PO₄^3– centers within one helix adopt the same configuration (ΛΛ shown in Fig. 1), but the complex is racemic (space group P2₁/c) with half of the supramolecules showing the opposite configurations. Although complex 1a lacks strict symmetry, the two ligands show very similar conformations and a non-distorted geometry would have one C₂ axis perpendicular to the P···P axis, but no C₂ axis through the two PO₄^3– centers. Each phosphate ion is coordinated by six urea groups (herein from two tris-urea units) through twelve hydrogen bonds (Fig. 1b and Table S2†). The two ligand strands are considerably twisted to form a large, partially open ‘cage’ in the middle of the helix, in which a TMA⁺ cation is encapsulated with N···P distances to the two phosphate ions of 5.10 and 5.38 Å, respectively (Scheme 1b). This ‘helicate cage’ is indeed similar to that formed in the A₂L₃ triple helicate with the methylene biphenylene-spaced bis–bis(urea) ligand,17b in which the ligands are also folded and compressed to accommodate a TMA⁺ guest (with N···P distances of 6.26 and 6.42 Å). The TMA⁺ ion is bound by electrostatic force and cation–π interactions with two phenyl rings of the p-xylylene spacer (purple-dashed lines, N···centroid distances are 4.426 Å to R1 and 4.340 Å to R2; Fig. 1c). The remaining five TMA⁺ ions serve as counter cations and are located around the helix (Fig. S12†).

Fig. 1. (a) Crystal structure of the double helicate [(TMA)⊂(PO₄)₂L₂]^5– (1a); (b) hydrogen bonds between a PO₄^3– ion and six urea units; (c) a TMA⁺ ion is encapsulated in the central cavity through cation–π interactions (purple-dashed lines; R1 and R2 denote the centroids of the aryl rings that are involved in cation–π interactions). Only an M enantiomer is shown; other counter cations and solvents are omitted for clarity.

In the assembly of anion-centered double helicates, the first one was constructed with polyguanidinium ligands and sulfate ions in solution.22 Later, several other double helicates based on halide anions were reported, e.g., the fluoride-directed helicate with neutral amide ligands23 and chloride-directed double helix with pyrrole-based oligomers.24 Very recently, Flood et al. reported the interconversion of single (foldamers) and double helicates glued by anions.25 However, complex 1a is the first double helicate with PO₄^3– anions.

The assembly of L and PO₄^3– ions and the encapsulation of TMA⁺ in solution was investigated by 1D and 2D NMR and ESI-MS. Compared to the free ligand, all the urea NH groups in complex 1a show significant downfield shifts (Δδ = 2.89–3.91 ppm, NMR of ligand L was measured in DMSO-d₆ for solubility reasons; Fig. 2b), corresponding to the coordination of urea to phosphate. A characteristic upfield shift of the H12 proton on the terminal nitrophenyl group (Δδ = –0.54 ppm; proton numbering, see Scheme 1a) is observed because of the strong shielding effects imposed in the helical structure. Interestingly, the two protons from one methylene of the spacer (H2) appear at different resonances due to their non-equivalent environments owing to the lack of symmetry in the double helicate (Fig. 1). This was also proved by 2D COSY which reveals the correlation between the two split peaks, indicating that they are from the same carbon (Fig. S25†).

Fig. 2. ¹H NMR spectra (400 MHz) of (a) ligand L in DMSO-d₆, 296 K; (b) double helicate 1a in CD₃CN, 296 K; (c) 1a in CD₃CN, 233 K; and (d) grid 2 in CD₃CN, 296 K. Signals of the two hydrogens of the methylene linker are shown in blue and turquoise, while that of trapped TMA⁺ ion is shown in red and protons of TEA⁺ in deep pink, respectively; \ represents solvent residue.

Notably, in the room-temperature NMR spectrum, there is only one peak for the two types of cations, i.e., encapsulated and peripheral TMA⁺ ions. This peak appears at an ‘averaged’ position of 2.80 ppm, only slightly upfield-shifted (Δδ = –0.28 ppm) relative to free TMA⁺ (3.08 ppm) in (TMA)₃PO₄, possibly due to the fast exchange between peripheral and trapped TMA⁺ guests at the NMR time scale. However, when the temperature is lowered to 233 K the peaks of ligands L split into two sets, indicating that the molecular motion is ‘frozen’ and the guest is ‘fastened’ (Fig. 2c). Meanwhile, the two different types of TMA⁺ cations become distinguishable, with the ‘missing’ signal of the trapped TMA⁺ (H_α) now showing a much more upfield shift at δ 1.64 ppm (Δδ = –1.44 ppm compared to the free state; see Fig. S9–S11† for variable-temperature ¹H NMR). In contrast, the protons of the five peripheral TMA⁺ ions (2.97 ppm) only shifted slightly upfield (Δδ = –0.11 ppm), revealing the lack of shielding on these ions, with an appropriate ratio (5 : 1) of the integral for the two types of TMA⁺ ions.

2D NMR spectra provide further evidence for the TMA⁺ cation encapsulation by the clear ¹H–¹H NOE (nuclear Overhauser effect) correlations between H_α of TMA⁺ and the phenylene protons (H1) of the linker of L (see Scheme 1 for proton numbering; Fig. S26†). The formation of a single host–guest species was confirmed by DOSY (diffusion-ordered nuclear magnetic resonance spectroscopy) (Fig. S27†). The HR ESI-MS spectrum of complex 1a also supports the formation of the A₂L₂ complex. Specifically, intense peaks at m/z 804.6570 (x = 3) and 1244.0363 (x = 4) corresponding to various A₂L₂(TMA)_n species, such as [(PO₄)₂L₂(TMA)_x]^(6–x)–, are seen in the spectrum (Fig. 4a).

Fig. 4. HR ESI-MS spectra of (a) complex (TMA)₅[(TMA)⊂(PO₄)₂L₂] (1a); and (b) complex (TEA)₁₀[(TEA)₂⊂(PO₄)₄L₄] (2).

Complex 2

By treating ligand L with 1.0 equiv. of (TEA)₃PO₄ that contains the larger TEA⁺ (tetraethylammonium) cation, the first anion-coordination-based grid complex, (TEA)₁₀[(TEA)₂⊂(PO₄)₄L₄] (2), was obtained. Its X-ray structure clearly shows a tetranuclear square conformation with four PO₄^3– ions positioned at the corners and four ligand molecules spanning the edges, two of which (opposite, magenta and bright green) are above the anion centers and the other two (yellow and cyan) are below (Fig. 3a). As required by a typical grid, the two diagonal PO₄^3– centers show the same chirality that is contrary to the other two corners, and the whole molecule (centrosymmetric space group I4₁/a) is achiral with alternating configurations (ΔΛΔΛ). The coordination sphere of PO₄^3– ions consists of six urea groups from two tris(urea) subunits that donate twelve hydrogen bonds to phosphate (Fig. 3c and Table S4†). Although known metal-directed M₄L₄ assemblies usually exhibit D₄ symmetries,20b this anion-directed A₄L₄ analogue exhibits S₄ symmetry, and the four symmetry-related PO₄^3– ions deviate slightly from co-planarity (by 0.58 Å), constituting a tilted square with four identical P···P distances of 11.95 Å and P···P···P angles of 89.5°.

Fig. 3. Crystal structure of the heterochiral (ΔΛΔΛ) grid complex 2 (TEA)₁₀[(TEA)₂⊂(PO₄)₄L₄]. (a) Top view; (b) side view; (c) hydrogen bonds formed between a PO₄^3– ion and six urea units. Some TEA⁺ cations are omitted for clarity (see Fig. S18† for the locations and interactions of these counter ions).

The negative charges of the anionic grid are balanced by twelve TEA⁺ counter-cations, which reside in four different environments. Notably, inside the grid, two TEA⁺ cations (type I) are readily accommodated, with one sitting over the other, an arrangement adopted possibly to facilitate cation···π interactions with the ‘aromatic box’ around them (Fig. 3b and S18a†). The two type-II TEA⁺ ions (distributed to four half-occupancy positions) are located at each vertex of the grid, forming hydrogen bonds with the terminal nitro groups (Fig. 3a and S18b†). Notably, each ligand strand as the ‘edge’ of grid 2 is folded to a helical conformation spanning ca. 1.5 turns. Within each ligand helix, a ‘half cavity’ is formed to encapsulate one of the third-type TEA⁺ cations (type III), which is surrounded by four aryl rings of the ‘host’ ligand and complemented by a terminal nitrophenyl plane from a neighboring ligand (Fig. S18a†). Finally, the remaining four TEA⁺ ions (type-IV) each cap a phosphate corner of the grid, forming intra- and inter-molecular interactions with ligand aromatic rings, thus linking two grids (Fig. S19†). All these cations, either internal or ‘peripheral’, may provide important templating or additional stabilizing effects for the grid structure, which also imply the possible guest inclusion ability of the voids in the grid complex 2.

In the ¹H NMR spectrum of grid 2, the urea NH groups also show large downfield shifts (Δδ = 2.93–4.14 ppm; Fig. 2d). The signal of the aryl H12 proton (Δδ = –0.49 ppm) undergoes an obvious upfield shift. The two protons from one CH₂ group of the spacer also split into two doublets due to their different chemical environments. Notably, although the crystal structure of 2 shows four different locations of TEA⁺ cations, they appear as only one set of signals in the NMR spectrum. Both of the CH₂ (H_α) and CH₃ (H_β) signals of TEA⁺ shift slightly upfield compared to that of the free (TEA)₃PO₄ (Δδ_α = –0.28 ppm and Δδ_β = –0.20 ppm) (Fig. 2d), and there is no significant change when the temperature is lowered (down to 233 K). This might reflect the facts that (a) there is a fast exchange among the protons at the four locations; and (b) these cations indeed form similar interactions with ligand L as revealed by the crystal structure. On the other hand, the HR ESI-MS spectrum confirms the composition of the A₄L₄ complex with the stoichiometry [(TEA)₉(PO₄)₄L₄]^3– (obsd 1851.9443 and cald 1851.8676, Fig. 4b). It is evident that complexes 1a and 2 display different NMR and ESI-MS signals, which can be used to distinguish the configuration of the double helicate or grid (see also the red dotted box in Fig. 5a).

Fig. 5. ¹H NMR spectra demonstrating the assembly of different phosphate salts with ligand L (400 MHz, CD₃CN, 296 K). (a) L with 1 equiv. of (TMA)₃PO₄, (TEA)₃PO₄, (TPA)₃PO₄ or (TBA)₃PO₄. (b) Stacking spectra of complexes 1a and 1c, and titration of 1c with (TMA)Cl ( denotes signals for TMA⁺, for TEA⁺, for TPA⁺, and for TBA⁺. The signal of trapped TMA⁺ is shown in red; \ represents solvent residue).

Assembly of the double helicates 1b and 1c

Considering the different outcomes in the two cases where only the counter cation varies, it appears that the counter cation (as the guest) plays a critical role in determining the supramolecular assembling process. To further elucidate the effect of the cation, we studied the assembly of L with one equivalent of (TPA)₃PO₄ or (TBA)₃PO₄ which contains the larger tetraalkylammonium homologues (TPA = tetrapropylammonium; TBA = tetrabutylammonium). In both cases, crystalline products (complexes 1b and 1c) were yielded. Although single crystals suitable for X-ray diffraction could not be obtained despite many attempts, the configuration of these complexes can be identified by a combination of ¹H NMR, HR ESI-MS, and theoretical calculations. The two products show ¹H NMR features that are similar to each other (Fig. 5a) but different from those of either complex 1a or 2, and the methylene H2 protons of the linker do not show any split, implying no significant guest interaction with the host. Their ESI-MS spectra are also similar, revealing a 2 : 2 anion-to-ligand ratio (Fig. S21 and S22†). Thus, we speculate that both complexes may have the double-stranded configuration, either double helicate or mesocate, with an ‘empty’ central cavity (solvent-filled) because either TPA or TBA is too large to fit. DFT calculations yield representative geometries for the helicate (Fig. S28†) and mesocate (Fig. S29†) assemblies in the absence of a guest. In the former case, optimization of the observed helicate geometry suggests that although the cavity becomes slightly smaller, the double helicate structure persists after the guest is removed.

Interconversion between the double helicate and grid motifs

Indeed, the guest-free complexes 1b/c provide an ideal model to examine the templating effect of the cation. For complex 1c, [(TBA)₆(PO₄)₂L₂], when (TMA)Cl was titrated in, the NMR spectrum gradually changed to that of the double helicate 1a. The methylene protons split into two peaks, and the encapsulation of TMA⁺ ions can be monitored. With the addition of less than 1 equiv. of TMA⁺ ions (per A₂L₂ unit) to complex 1c, the methyl signal H_α showed a dramatic upfield resonance at 1.74 ppm (Δδ = –1.34 ppm compared to free TMA⁺), indicating the encapsulation of TMA⁺. However, this signal shifted back downfield when more than 1 equiv. of TMA⁺ cations was added (Fig. 5b), which was caused by the fast exchange of the encapsulated and excess peripheral TMA⁺ cations, as in the case of complex 1a. Finally, with more TMA⁺ cations (6 equiv.), the titration reached saturation and the spectrum is almost the same as that of 1a.

Inspired by the NMR titration results, which indicated that the initially ‘empty’ double helicate (TBA)₆[(PO₄)₂L₂] (1c) can encapsulate one TMA⁺ cation (and thus displace one TBA⁺ ion), we were able to isolate single crystals of the double helicate complex 1d, (TBA)₅[(TMA)⊂(PO₄)₂L₂] (Fig. S16 and Table S3†) from L and a mixture of phosphate salts of TMA⁺ and TBA⁺ in a 1 : 5 ratio. The main structure [(TMA)⊂(PO₄)₂L₂]^5– in 1d is essentially the same as that in complex 1a, and the difference lies only in the peripheral counter cations, TBA⁺ in 1dversus TMA⁺ in 1a. In addition, HR ESI-MS also shows the signal of [(TBA)₃(TMA)(PO₄)₂L₂]^2– (m/z obsd 1496.3825 versus calcd 1496.2675, Fig. S20†).

Subsequently, the conversion of double helicate to grid motifs was studied using ¹H NMR titration (Fig. 6a) by introducing the TEA⁺ ions into helicate 1c. With the addition of 1 equiv. of (TEA)Cl to a solution of 1c in CD₃CN, the methylene protons H2 of the spacer split to two peaks, a characteristic feature of grid 2. In the aromatic area, the H1 signal shifts upfield due to encapsulation of a TEA⁺ ion with the conformational change. Meanwhile, both peaks of TEA⁺ appear in the upfield region (without peripheral TEA⁺ ions) compared to those of complex 2 but shift downfield gradually with more added TEA⁺ ions until becoming identical to complex 2. On the other hand, the reverse process, i.e. transformation of the grid structure 2 to double helicate (1a) is also readily realized upon addition of the TMA⁺ template (Fig. 6b). With the encapsulation of TMA⁺, the aryl H1 signal of the spacer shifts slightly upfield until reaching the same as complex 1a, and H4 and H5 peaks shift downfield due to the deshielding effect after encapsulation of the TMA⁺.

Fig. 6. Partial ¹H NMR titration spectra demonstrating the interconversion of different complexes (400 MHz, CD₃CN, 296 K). (a) From ‘empty’ double helicate to grid 2 by titrating 1c with (TEA)Cl. (b) From grid 2 to double helicate 1a by titrating complex 2 with (TMA)Cl ( denotes signals for TMA⁺, for TEA⁺, and for TBA⁺. Signals of trapped TEA⁺ ions are shown in red; \ represents solvent residue).

Thus, we have observed the conversion not only from complexes 1b/c to 1a, namely from ‘empty’ to the TMA⁺-filled double helicate, but also from 1b/c to grid 2 under the modulation of TEA⁺ cations. Conversely, due to the strong binding of the helical cage to TMA⁺, transformation of grid 2 to double helicate 1a is also very feasible. Such template-induced supramolecular transformations have also been reported for metal-coordination driven assembly, in which the counter-anions are responsible for the conformational changes.2c,6c,7d,9a,b

Conclusions

We reported the design and synthesis of a C₂-symmetric bis–tris(urea) ligand with a p-xylylene linker, which is assembled with a phosphate anion to form the first anion coordination-based 2 × 2 grid (A₄L₄) and a double helicate (A₂L₂) featuring an open cavity similar to DNA grooves. Moreover, facile interconversion between these two motifs can be realized simply by modulation with the counter-cation under ambient conditions, which is also complementary to anion-controlled structural interconversion. This work reveals the power of anion coordination in supramolecular assembly. Such aniono-supramolecular systems could be a new class of architectures for promising applications in many fields. Construction of more complex structures and exploration of their functionality is currently underway.

Conflicts of interest

The authors declare no competing financial interest.