Chemoresponsive alternating supramolecular copolymers created from heterocomplementary calix[4]pyrroles

Abstract

The importance of noncovalent interactions in the realm of biological materials continues to inspire efforts to create artificial supramolecular polymeric architectures. These types of self-assembled materials hold great promise as environmentally stimuli-responsive materials because they are capable of adjusting their various structural parameters, such as chain length, architecture, conformation, and dynamics, to new surrounding environments upon exposure to appropriate external stimuli. Nevertheless, in spite of considerable advances in the area of responsive materials, it has proved challenging to create synthetic self-assembled materials that respond to highly disparate analytes and whose environmentally induced changes in structure can be followed directly through both various spectroscopic and X-ray diffraction analyses. Herein, we report a new set of artificial self-assembled materials obtained by simply mixing two appropriately chosen, heterocomplementary macrocyclic receptors, namely a tetrathiafulvalene-functionalized calix[4]pyrrole and a bis(dinitrophenyl)-meso-substituted calix[4]pyrrole. The resulting polymeric materials, stabilized by combination of donor–acceptor and hydrogen bonding interactions, undergo dynamic, reversible dual guest-dependent structural transformations upon exposure to two very different types of external chemical inputs, namely chloride anion and trinitrobenzene. The structure and dynamics of the copolymers and their analyte-dependent responsive behavior was established via single crystal X-ray crystallography, SEM, heterocomplementary isodesmic analysis, 1- and 2D NMR, and dynamic light scattering spectroscopies. Our results demonstrate the benefit of using designed heterocomplementary interactions of two functional macrocyclic receptors to create synthetic, self-assembled materials for the development of “smart” sensory materials that mimic the key biological attributes of multianalyte recognition and substrate-dependent multisignaling.

In recent years, considerable effort has been devoted to the development of artificial supramolecular polymeric materials (1–20). Much of the interest in these kinds of systems stems from an appreciation that they act as rudimentary models for some of the most fundamental, yet complex systems, found in nature. Living systems typically exploit multiple noncovalent interactions to drive the formation and stabilization of a wide variety of complex polymeric architectures for the implementation of function, codifying diversity, and imparting complexity. Because of their stabilization, in part by multiple weak forces, including hydrogen bonds, metal-ligand interactions, and donor–acceptor interactions, many of these architectures undergo considerable modification in their structure and function upon application of some external stress or exposure to a specific chemical trigger, as illustrated by various denaturation and renaturation processes involving DNA (21), proteins (22), and lipid bilayers (23). Although varying widely in terms of the specifics, as a general rule, the functional response to activating molecular stimuli seen in biological systems can be traced to substrate-induced structural changes involving large, multicomponent systems built up from smaller fragments through self-assembly.

These same weak noncovalent interactions also provide the basis for the fast-evolving field of self-assembled and responsive materials (3–20). Here, important advances have been made of late, including the development of self-healing materials (18–20) and stimuli-responsive supramolecular polymers (3–10). Although a variety of strategies have been advanced for the construction of supramolecular polymeric systems, to date it has not proved possible to create artificial, stimulus-responsive, synthetic materials that reflect the full complexity of natural functional systems. Particularly challenging has been the creation of multicomponent self-assembled systems that respond in an orthogonal fashion to two or more guests and which give rise to easy-to-discern guest-dependent multiple readout signals. Even more elusive are responsive polymeric materials that have been fully characterized at the level of atomic resolution via single crystal X-ray diffraction analysis

We now report a class of structurally characterized heterocomplementary, self-assembled materials obtained from the pairing of electron-rich tetrathiafulvalene-calix[4]pyrroles (TTF-C4Ps 1, 2, and 3) (24, 25) and electron poor cis/trans-bis(dinitrophenyl)-calix[4]-pyrroles (DNP-C4Ps 4 and 5) (SI Appendix). As detailed below, these macromolecular materials (e.g., 6) are stabilized via a combination of oriented hydrogen bonding and pi–pi donor–acceptor interactions. They act as “intelligent” chemoresponsive materials, displaying distinctive changes in structure, color, electrochemical and spectroscopic features when exposed to appropriate guests, specifically chloride anion and the test nitroaromatic explosive, trinitrobenzene (TNB). Both the initial supramolecular polymerization process and the ensuing analyte-induced changes were followed in solution by UV-visible (UV-vis) and ¹H-, NOESY-, and diffusion ordered nuclear magnetic resonance (DOSY-NMR) spectroscopies, as well as dynamic light scattering (DLS). The associated structures were also determined in the solid state via single crystal X-ray diffraction analyses (Figs. 1 and 2).

Fig. 1.

Schematic representation of the self-association between the TTF-C4P monomers 1, 2, and 3 and the heterocomplementary DNP-C4P monomers 4 and 5. Also shown is the chemoresponsive behavior seen upon the addition of either TEACl or TNB.

Fig. 2.

Solid-state analyses of both alternating supramolecular polymeric arrays made up from the combinations of TTF-C4Ps and DNP-C4Ps and the TEACl- and TNB-containing forms of these same calix[4]pyrroles (1, 4, and 5): (A) SEM image of 6a, [1 + 4]. (B) Crystal structure of 6a ([1 + 4]). (C) SEM image of 6d, [2 + 5]. (D) Crystal structures of 6d ([2 + 5]). (E) Crystal structures of 1•TEACl, (F) 4•TEACl, (G) 5•TEACl, and (H) 1•(TNB)₂. Inset boxes show photographs of the datum crystals.

TTF-C4Ps, like other calix[4]pyrroles, can adopt several conformations, including the limiting cone and 1,3-alternate forms. In previous work, we found that these macrocycles would bind planar, electron-deficient nitroaromatics, including TNB, with a 1∶2 stoichiometry in their 1,3-alternate forms in the absence of anions (24, 25). In contrast, the corresponding anion-stabilized cone forms would interact with larger electron-deficient species, as evidenced by the anion-triggered binding of C₆₀ (26) or the stabilization of a long-lived electron transfer state as the result of supramolecular capsule formation (27). The fact that we could stabilize such discrete supramolecular structures led us to consider that doubly functionalized electron-deficient guest species, such as 4 and 5, could be utilized as ditopic macrocyclic building blocks to create self-assembled polymeric arrays. We also postulated that the resulting self-assembled systems would prove responsive to multiple chemical stimuli. Specifically, we predicted that the noncovalent materials produced by the pairing of 4 and 5 with 1–3 would undergo a loss of structural integrity when exposed to either (i) more electron-deficient species, such as TNB (which would compete with 4 or 5 for 1–3), or (ii) a strongly bound anionic substrate, such as chloride (which would trigger conversion to the cone conformation). This proposed dual analyte-dependent behavior is shown schematically in Fig. 1.

Results and Discussion

Preparation of Alternating Supramolecular Copoplymers.

To test the hypothesis shown in Fig. 1, two diasteromeric isomers 4 and 5, (DNP-C4Ps) were prepared via a two-step synthetic procedure that passed through the corresponding 1,3-dinitrophenyl substituted dipyrromethane (SI Appendix, Section 2). In the case of both 4 and 5 and their dipyrromethane precursors, it was believed that the dinitrophenyl (DNP) groups would allow interaction with TTF-C4Ps, such as 1, 2, and 3, via a combination of TTF-DNP donor–acceptor interactions and pyrrole NH-nitro group hydrogen bonds. In fact, this association was observed for both the dipyrromethane precursor (SI Appendix, Fig. S13) and the ditopic systems, 4 and 5 (see below). Initial evidence that the TTF-C4Ps 1, 2, and 3 could complex the DNP-C4Ps 4 and 5 came from the observation of an immediate color-change from yellow to brown when various combinations of these donors (1, 2, and 3) and acceptors (4 and 5) were mixed in equal proportions in chloroform. A broad low-energy absorption (415–750 nm) was also seen in the UV-vis absorption spectrum of this material (SI Appendix, Fig. S30).

Solid-State Characterization of Alternating Supramolecular Copolymers 6.

In order to characterize the solid-state products obtained as the result of these mixing experiments, pentane was slowly diffused into the initial chloroform solutions. In all cases, this crystallization procedure afforded long, brownish needle-shaped materials except for the combination of 3 and 5 (SI Appendix, Sections 3-1). The nature of these solids differed significantly in terms of both morphology and color from the starting materials 1–5 when subject to similar pentane-chloroform crystallization procedures. Detailed SEM studies of the materials 6a–e obtained from the mixtures of 1–3 and either 4 or 5 revealed the formation of well-defined, long needle-like fibers from the five combinations. Collectively, these data provide support for the inference that the solid material was obtained as the result of a highly directional 1D crystal growing process (see Fig. 2 A and C, and SI Appendix, Figs. S10–S12).

In the case of 6a and 6d obtained from 1 + 4 and 2 + 5, respectively, diffraction-quality single crystals could be cut from bundles of the long, thin needles formed as the result of the pentane-chloroform crystallization process. The corresponding solid-state structures, shown in Fig. 2 B and D, provided unequivocal evidence for the formation of linear, alternating supramolecular copolymers. On the basis of the structural parameters, it is inferred that the self-assembled materials in question (i.e., 6a and 6d) are built up from the corresponding TTF-C4P and DNP-C4P precursors as the result of directional hydrogen bond and donor–acceptor interactions. These interactions involve (i) the TTF-pyrrolic NH protons and the oxygen atoms of the DNP nitrogroups (ca. 3.05 Å) and (ii) the electron-rich TTF and electron-deficient DNP groups (ca. 3.26–3.60 Å between TTF-plane and the nitrogroups). Although produced from different building blocks, it is noteworthy that the structures of 6a and 6d are similar; both are characterized by a linear alternating sequence of ditopic constituents and both contain calix[4]pyrroles in their limiting 1,3-alternate conformations. The resulting 1D order presumably lies at the origin of the high crystallinity seen for these polymeric materials, especially those built up from the annulated TTF-C4Ps 1 and 2. This crystallinity precluded efforts to determine viscosities for the ensembles. However, in the case of several representative systems the self-assembled materials could be characterized fully in solution using spectroscopic methods as detailed below.

Solution-state Analyses of Heterocomplementary Binding Interactions.

The self-association between TTF-C4Ps 1 and 3 and the DNP-C4Ps 4 and 5 was also studied in chloroform solution using electronic absorption, ¹H-NMR, NOESY, DOSY, and DLS spectroscopies. Unfortunately, the systems formed from 2, although characterized in the solid state in the case of ensemble 6d (see above), proved too insoluble to permit detailed spectroscopic analysis.

Initial support for the proposed intercomponent complexation came from NOESY NMR studies carried out in CDCl₃; in the case of the putative self-associated system 6a, the presence of cross-peaks corresponding to close, noncovalent contacts between the two components 1 and 4 were seen (Fig. 3A). The presence of these cross-peaks is taken as evidence that these two components interact in solution. However, their observation does not per se provide evidence of the preferred binding stoichiometry or the average length of the aggregate (i.e., number of subunits in the presumed oligomeric array) that pertains in solution.

Fig. 3.

Solution-state analyses of the heterocomplementary binding interactions between TTF-C4P and DNP-C4P monomers. (A) An expanded region of a 2D NOESY NMR spectrum of 6a, [1 + 4], showing the NOE between methyl CH₃ (H_d) of 1 and dinitrophenyl protons of 4 (H_a and H_b). (B) Continuous variation plots for the chloroform mixture of [1 + 4] (black square) and [1 + 5] (red square) and their simulated nonlinear curve fits. (C) Estimated number average aggregation (N) of 6a–f as a function of total concentration as calculated using a heterocomplementary isodesmic binding model. Note that for ensembles 6c [2 + 4] and 6d [2 + 5] the curves at concentrations above 0.4 mM of theoretical interest only because this concentration represents the solubility limit for these systems. The inset gives the associated K values. (D) Dynamic light scattering spectra of 6a, 6b, 6e, and 6f recorded at a total monomer concentration of 0.6 mM. (E) Changes in diffusion coefficients of 6a seen upon varying the total monomer concentration (red squares) and upon the addition of external guest species (TNB and TEACl, black and green squares, respectively) as determined from DOSY-NMR spectroscopic analyses.

In accord with the proposed heterocomplementary interactions between the TTF-C4P and DNP-C4P building blocks, the stoichiometry of mixing proved critical. Because an excess of either one of the components should act as a chain growth stopper, a 1∶1 ratio was expected to be most efficient in terms of stabilizing polymeric aggregates. Support for this supposition came from continuous variation or so-called Job plots involving mixtures of TTF-C4P 1 and either DNP-C4P 4 or 5. Specifically, by monitoring the change in absorbance at 450 nm as a function of relative subunit concentration, a symmetric curve was obtained that displayed a maximum at a mole fraction, x = 0.5 in both cases (Fig. 3B). This result was taken as evidence that the best associations and the longest chains would pertain when the electron-rich and electron-deficient components of this study were mixed in equal proportions.

As expected for self-associated materials (28, 29), the molar extinction coefficients for equimolar mixtures of 1–3 with 4 or 5 recorded in chloroform were found to depend on the total concentration. Moreover, plots of extinction coefficient vs. concentration for all cases yielded hyperbolic curves (SI Appendix, Fig. S25). Nonlinear curve-fitting analyses of these plots were carried out using an isodesmic equal K model (28), modified to account for the heterocomplementary binding motifs involved (SI Appendix, Section 4). This approach was considered justified by the highly symmetric nature of the Job plots and the self-consistent nature of the treatment. It gave calculated association constants (K) for each individual system (Fig. 3C) and allowed for intercomparisons between the self-assembled materials produced by various permutations of 1–3 and 4 or 5.

Although plagued by low solubility (see above), the systems derived from 2 (i.e., 6c and 6d) were found to display the strongest interactions as inferred from the isodesmic analyses, followed by those produced from 1 and 3, as reflected in the calculated K values (see Fig. 3C). The K value for the structurally characterized, self-associated systems 6d and 6a were 7.9 × 10⁵ M^-1 and 5.9 × 10⁴ M^-1, respectively. For the soluble system, 6a, the average length of the oligomer (N) was estimated on this basis to be ca. 21 at a total concentration of 1 + 4 = 14 mM; this N value fell to ca. 7 when the concentration of 1 + 4 was reduced to 1.4 mM (N_14 mM/N_1.4 mM ≈ 3 on a subunit basis).

Further confirmation that supramolecular aggregates of inferred structure 6a are formed in the case of the more soluble pair 1 + 4 came from DOSY-NMR and DLS analyses (Fig. 3 D and E). In the first of these studies, the diffusion constants of equimolar mixtures of 1 and 4 were recorded at two limiting total monomer concentrations, namely 1.4 and 14 mM, respectively. As reflected in Fig. 3E, a substantial decrease in the diffusion constant from ca. 30 × 10^-6 to ca. 10 × 10^-6 cm²/s (D_1.4 mM/D_14 mM ≈ 3) was seen on moving to the higher concentration. This result is consistent with the formation of larger self-assembled aggregates at higher total monomer concentrations. DLS analyses (30) also provided support for the conclusion that larger aggregates are formed when 1 and either 4 or 5 are mixed, as compared to the combination involving 3 (Fig. 3D). Unfortunately, solubility considerations precluded analogous studies being carried out with 2.

Chloride Responsive Behavior of Supramolecular Polymers 6a and 6b.

As observed with other calix[4]pyrroles, both TTF-C4Ps 1–3 and DNP-C4Ps 4 and 5 bind the chloride anion well in chloroform solution, as reflected in K values of 2.3 × 10⁷, 5.6 × 10³, and 2.0 × 10⁵ M^-1 for the interaction of tetraethylammonium chloride (TEACl) with 1, 4, and 5, respectively (SI Appendix, Section 4-3). The anion binding interaction induces the formation of a cone conformation, which is not able to complex effectively flat, electron-deficient species, such as the DNP residues present in 4 and 5. Polymers of general structure 6 were thus expected to respond to the chloride anion and undergo deaggregation as a consequence. This hypothesis was explicitly tested in the case of 6a and 6b.

In the case of 6a, addition of two equivalents of TEACl to the original chloroform-d₁ solution resulted in the appearance of two sets of signals in the ¹H NMR spectrum, corresponding to the chloride anion complexes of the constituent calixpyrroles (i.e., [1•Cl]^- and [4•Cl]^-), respectively (SI Appendix, Fig. S35). A loss in intraresidue cross-peak intensities was also seen in the NOESY NMR spectrum (SI Appendix, Fig. S40), as were larger diffusion coefficients in the DOSY-NMR spectrum (Fig. 3E and SI Appendix, Fig. S37). Anion complexation by 1 and 4 also resulted in changes in the optical and electrochemical properties of 6a as reflected in analyte-dependent changes in the UV-vis spectra (SI Appendix, Figs. S31 and 32) and cyclicvoltammograms (SI Appendix, Fig. S41). Taken in concert, these spectroscopic changes are consistent with deaggregation and competitive chloride anion binding, as per the design expectations illustrated in Fig. 1.

Further evidence for deaggregation came from solid-state structural analyses. When chloroform solutions of 6a or 6b were treated with 2 equiv of TEACl, two sets of distinctive crystals were formed using the same crystallization method (pentane-chloroform) used to produce the self-assembled materials. Single crystal X-ray diffraction analyses of these crystals revealed the TEA⁺-bound forms of [1•Cl]^- and either [4•Cl]^- or [5•Cl]^-, respectively (Fig. 2 E–G). As expected given the competitive nature of the interactions involved, washing the original chloroform solutions of 1 + 4 and TEACl with water served to restore the original brown color. Such findings provide further support for the environment-dependent, analyte-triggered equilibrium shown in Fig. 1.

TNB Responsive Behavior of Supramolecular Polymer 6a.

The external addition of TNB (2 equiv) into an equimolar mixture of 1 and 4 in chloroform also resulted in a breakup of the corresponding self-assembled array (6a). In this case, the deaggregation process was driven by competition for the donor sites afforded by the 1,3-alternate form of TTF-C4P 1 and is consistent with TNB functioning as a better acceptor (more electron deficient) than the DNP residues present in 4. As above, this stimulus-induced response was monitored by ¹H-NMR, NOESY, and DOSY spectroscopies (seeFig. 3E and SI Appendix, Section 4-6). It was also reflected in color change from brown to dark blue (see Fig. 1) and a corresponding increase in absorptivity at a λ_max of 670 nm (SI Appendix, Fig. S31). TNB complexation by monomer 1, a component originally present in 6a, was also reflected in a large positive shift of the first oxidation potential () for 1, along with significant changes in the redox behavior of the added TNB (SI Appendix, Fig. S42).

When either of the self-associated materials 6a or 6d in chloroform were treated with 2 equiv of TNB and then subject to pentane diffusion in analogy to what was done in the case of TEACl, two sets of crystals were obtained. On the basis of X-ray crystallographic analyses, these proved to be 1•(TNB)₂ (Fig. 1H) and the substrate-free, trans-2,4,6,-trinitrophenyl-C4P 4. These solid-state findings thus provide additional support for the solution-state analyses described above.

Conclusion

Although the addition of both chloride anion and TNB leads to a chemical stimulus-triggered change in polymer structure, it is important to appreciate that these two analytes operate via different mechanisms and give rise to different products. Their respective addition also produces different colors and gives rise to distinctive electrochemical responses. These results lead us to suggest that appropriately designed synthetic materials, such as the self-assembled systems of this report, can now be used to distinguish between highly disparate classes of guests. Such discrimination is manifest in terms of a different downstream response (i.e., changes in structure, chemical form, optical and electrochemical features), just as it is for many complex biological ensembles. The ability to follow this response both in solution and in the solid state is a unique feature of the system described herein, and it is expected to further our understanding of how multicomponent, stimulus-responsive systems, both synthetic and natural, operate at the molecular level.

Materials and Methods

Compounds 1, 2, and 3 (TTF-C4Ps) were prepared according to literature procedures (19, 20), whereas compounds 4 and 5 (DNP-C4P) were synthesized via the trifluoroacetic acid catalyzed condensation reaction between acetone and 1-(3,5-dinitrophenyl)ethanone. All supramolecular polymer fibers (6a–e) and single crystals of 6a and 6d were prepared from the slow diffusion of pentane into equimolar mixtures of the TTF-C4P (1, 2, and 3) and DNP-C4P (4 and 5) monomers in chloroform solution. Similarly, single crystals of 1•TEACl, 4•TEACl, 5•TEACl, 1•(TNB)₂, and 2•(TNB)₂ were prepared from pentane diffusion of chloroform solution of 6 containing 2 equivalents of TEACl or 4 equivalents of TNB.

For estimation of degree of polymerization as a function of concentration profile, the equation , (K, C_T, ϵ₁, and ϵ_a are the association constant, total concentration of monomers, the extinction coefficients for the monomers, and the aggregate species, respectively) was used for the calculation of K from the nonlinear regression analysis of plots of the extinction coefficients vs. concentration of 6. Based on these calculated K values, number-average aggregate sizes (N) were calculated according to the equation . Full theoretical procedures for deriving the above equations and experimental details of synthesis and characterization for all compounds with instrumentation are included in the SI Appendix.