Mechanistic Insights into the Self‐Assembly of an Acid‐Sensitive Photoresponsive Supramolecular Polymer

Abstract

The supramolecular polymerization of an acid‐sensitive pyridyl‐based ligand (L₁) bearing a photoresponsive azobenzene moiety was elucidated by mechanistic studies. Addition of trifluoroacetic acid (TFA) led to the transformation of the antiparallel H‐bonded fibers of L₁ in methylcyclohexane into superhelical braid‐like fibers stabilized by H‐bonding of parallel‐stacked monomer units. Interestingly, L₁ dimers held together by unconventional pyridine–TFA N⋅⋅⋅H⋅⋅⋅O bridges represent the main structural elements of the assembly. UV‐light irradiation caused a strain‐driven disassembly and subsequent aggregate reconstruction, which ultimately led to short fibers. The results allowed to understand the mechanism of mutual influence of acid and light stimuli on supramolecular polymerization processes, thus opening up new possibilities to design advanced stimuli‐triggered supramolecular systems.

A look into the self‐assembly: Detailed mechanistic insights into the supramolecular polymerization of an acid‐sensitive photoresponsive pyridyl‐based ligand are reported. Addition of acid (TFA) caused the transformation of long twisted fibers into superhelical braid‐like assemblies stabilized by aromatic, H‐bonding, and van der Waals interactions through a dimerization event involving two unconventional pyridine–TFA N⋅⋅⋅H⋅⋅⋅O bridges. Photoirradiation led to a strain‐driven disassembly of the superhelical fibers and subsequent attenuated growth into short fibers after reconstruction.

Introduction

Supramolecular functional materials derived from discrete π‐systems have attracted considerable attention during the past few decades, primarily due to their promising optoelectronic1 and biomedical applications.2 These smart materials held together through multiple noncovalent interactions have the advantage of reversible changes in the physical properties as a response towards various stimuli, including temperature,3 light,4 mechanical stress,5 ultrasounds,6 metal ions,7 redox reactions8 and pH.9 Successful candidates are often used as sensors,10 switches,11 as well as for drug delivery.12 In particular, light‐responsive organogelators derived from azobenzenes,13 spiropyrans14 and diaryethenes15 have been widely exploited in the field of self‐assembled materials due to their potential application in the area of storage and secret documentation.16 The trans‐to‐cis photoisomerization of the azobenzene groups has been used to reversibly control gel–sol transitions, which eventually has produced smart photoresponsive materials.17 In addition to the significant development in the field of light‐responsive organogelators, more advanced multistimuli‐responsive gelators based on azobenzene have been recently reported.18, 19 However, detailed mechanistic investigations of their triggered self‐assembly remain scarce.

Recently, we have reported the reciprocal impact of light irradiation and metal coordination on the hierarchical self‐assembly of an azo‐containing pyridyl ligand (molecule L₁ in Scheme 1).20 The ability of the pyridyl ligand to bind metal ions, such as Pd^II, was investigated in terms of its influence on the self‐assembly and photoresponsive behavior. However, the pyridyl group can be also exploited as an acid‐sensitive unit in order to effectively tune self‐assembly processes in solution. Here, we examine the interplay of acid and light irradiation in the supramolecular polymerization of L₁ (Scheme 1). Interestingly, we found that addition of acid to the fiber‐like assemblies of L₁ enhances the aggregation and leads to the formation of unique superhelical structures through stacking of acid‐bridged L₁‐dimers. Furthermore, we found that the light irradiation could cause disassembly of the superhelical fibers and their reconstruction into thin fibers through a different pathway associated with a rapid cis–trans isomerization under isothermal conditions.

Scheme 1.

Molecular structure of L₁ and schematic representation of its self‐assembly behavior as a response towards acid and UV light. The superhelical fibers (right, bottom) were optimized through dispersion‐corrected PM6 semiempirical calculations (nonpolar hydrogens were omitted for clarity).

Results and Discussion

Self‐assembly and gelation of L₁

Gelation studies of L₁ were performed in different organic solvents and the majority of them were immobilized upon addition of L₁ in excess. Pictures of all gels under daylight are shown in Figure S1 (Supporting Information), whereas Table S1 provides information about their critical gelation concentration (CGC) values in various solvents. A minimum CGC of 2.31 mg mL⁻¹ in isooctane and a maximum of 16.36 mg mL⁻¹ in n‐butanol were marked as the lower and upper benchmarks. From these values, one can clearly notice that hydrophobic solvents readily and most effectively gelate L₁ (Table S1, Supporting Information). The CGC values decrease upon increasing the n‐hydrocarbon chain length of the solvents, confirming the role of hydrophobic interactions.

Interestingly, gels in toluene cannot be obtained even at very high concentrations, which could be explained in terms of attenuated aromatic interactions. Moreover, n‐alcohols can also induce gelation, although with higher CGC values compared to analogous n‐hydrocarbons. These results suggest a strong contribution of H‐bonding to the gelation process in hydrocarbon solvents, whereas in n‐alcohols these interactions are weakened. On this basis, we can conclude that H‐bonding as well as aromatic and hydrophobic interactions are the primary forces responsible for gel formation. The morphology of the xerogels formed in selected solvents [isooctane and methylcyclohexane (MCH), see Figure 1] was examined by atomic force microscopy (AFM) on highly oriented pyrolytic graphite (HOPG). For both samples, densely packed networks of entangled fibers could be observed. Additionally, scanning electron microscopy (SEM) studies on glass substrate confirmed the fibrillar morphology of the xerogels (Figure S2, Supporting Information).

Figure 1.

Surface morphology of xerogels of L₁ in isooctane (a) and MCH (b) visualized by AFM on HOPG.

Considering that MCH yielded stable gels and is commonly used for aggregation studies, we selected this solvent as ideal medium for gelation as well as pH‐ and light‐responsive studies. Figure 2 a shows the variable‐temperature (VT) ¹H NMR experiments at 11.7 mm obtained by cooling a solution of L₁ in MCH‐d₁₄ from a molecularly dissolved state (363 K) to a gel state (298 K) in steps of 2.5 K. The spectra show noticeable chemical shifts upon decreasing temperature. For instance, although the signals of the amide protons H_e and those of protons H_d are downfield shifted, the proton signals of the outer aromatic rings (H_f) undergo a shielding effect upon decreasing temperature (Figure 2 a). Alternatively, pyridyl protons H_a and H_b and protons H_c from the central aromatic rings are barely sensitive to temperature (Figure 2 a). In particular, the strong downfield shift (≈7.55 to 8.95 ppm) observed for the amide protons (H_e) upon decreasing temperature (Figure 2 a,b) confirms the active involvement of H‐bonding in the aggregation process. A comparably smaller downfield shift observed also for H_d can be due the close proximity of ‐C=O group. In addition, the observed shifts for protons H_a–d are in perfect agreement with the behavior of L₁ in more diluted solutions (10⁻⁴  m) and indicate an anti‐parallel organization of the molecules of L₁ in the MCH gel (Figure 2 b).20 The changes in the NMR chemical shifts of some of these protons, that is, H_c and H_d, were monitored upon temperature variation to determine the aggregation mechanism and the corresponding thermodynamic parameters.

Figure 2.

a) VT ¹H NMR of L₁ at 11.7 mm in MCH‐d ₁₄, b) antiparallel L₁ hexamer optimized by dispersion‐corrected PM6 calculations. c) plot of α _agg vs. T and fit to the isodesmic model (α _agg was calculated by monitoring the shifts of H_d).

Plotting the fraction of aggregated species (α _agg) against temperature afforded a sigmoidal curve, suggesting that the self‐assembly of L₁ occurs in a non‐cooperative manner. Application of the isodesmic model21 affords good fits (Figure 2 c) whereas the corresponding Van't Hoff fit shows clear linearity (Figure S3, Supporting Information), demonstrating the suitability of the aggregation model to our data. The self‐assembly process is enthalpy‐driven (−63.62 kJ mol⁻¹) with an association constant (K) of 585 m ⁻¹. The melting temperature (T_m) for the assembly process is 330 K (Table S2, Supporting Information). This relatively low K value also matches well with the previously observed lack of aggregation of L₁ in more diluted MCH solutions (0.1 mm) at room temperature.

Influence of addition of acid on the self‐assembly of L₁

The presence of a basic pyridine moiety in L₁ endows our system with stimuli‐responsiveness towards acids. Initially, a gel (11.7 mm) in MCH was prepared and its sol‐to‐gel reversibility was tested for several cycles (Figure 3). Addition of 2 μL of trifluoroacetic acid (TFA, 1 equiv) leads to the transformation of the orange‐colored opaque gel into a clear red‐colored solution only few seconds after TFA addition. These findings strongly suggest the protonation of the pyridine groups of L₁ upon TFA addition and the subsequent cleavage of the gel network, most likely as a result of electrostatic repulsion of the positive charges and decrease in solubility of the monomer units. Notably, the gel can be regenerated within 5 minutes after addition of 1 equivalent of triethylamine (Et₃N) (Figure 3) due to the deprotonation of the pyridyl groups. The reversibility of the sol‐to‐gel transformation could be reproduced for various cycles of acid/base addition without signs of decomposition.

Figure 3.

Pictures showing the reversible sol‐to‐gel transformation of L₁ in MCH (11.7 mm) on heating and cooling and addition of TFA and Et₃N.

To gain closer insights into the impact of acid addition on the aggregation behavior, an MCH solution of L₁ under more diluted conditions (5×10⁻⁴  m) was titrated against TFA and monitored by UV/Vis absorption spectroscopy. A significant redshift from 362 to 390 nm with broadening in the absorption spectrum was observed upon addition of 0.1 to 1 equivalents of TFA to the L₁ solution, which can be assigned to the protonation of the pyridyl group of L₁. No detectable emission was observed for L₁ before or after addition of TFA. To rule out a partial aggregation of L₁ prior to TFA addition (L₁ was previously observed to aggregate slowly over time at 5×10⁻⁴  m in MCH), we titrated L₁ with TFA in a good solvent (chloroform), in which both species are soluble and no aggregation of L₁ takes place. Similarly to MCH, the formation of a red‐colored solution upon addition of 1 equiv TFA supports the protonation of the pyridyl group of molecularly dissolved L₁ (Figure 4 a, inset). Subsequently, the solvent (chloroform) was removed to obtain the L₁:TFA species as a red powder. Finally, to examine the impact of TFA addition on the self‐assembly, the powder was redissolved in MCH to induce aggregation. This equimolar mixture of L₁ and TFA in MCH was subsequently heated to 363 K to ensure the formation of a monomer state and subjected to a VT UV/Vis experiment using a cooling rate of 1 K min⁻¹ (Figure 4 b). Upon cooling down to 303 K, a 10 nm redshift in the absorption spectrum was observed (377 to 387 nm). This behavior can be ascribed to a J‐type aggregate formation, which has been previously observed for related acid‐sensitive pyridine‐based two‐component systems.22 Additionally, further cooling to 298 K induces the formation of higher order aggregates in solution, which is also visible with the naked eye and evidenced by a significant decrease of the absorption band (Figure 4 b).

Figure 4.

a) Absorption changes of L₁ (5×10⁻⁴  m) in CHCl₃ upon addition of increasing amounts of TFA (inset: photographs showing the color change upon addition of 1 equiv of TFA). b) VT UV/Vis (0.5 mm, MCH) and c) VT ¹H NMR experiments of L₁:TFA complex between 363 and 298 K (0.5 mm, MCH‐d ₁₄).

VT ¹H NMR measurements under identical conditions showed a significant deshielding (8.11 to 9.29 ppm) of the amide N−H protons on cooling, indicative of strong H‐bonding (Figure 4 c). Alternatively, all other aromatic signals were shielded upon cooling due to their active involvement in π–π stacking. These shifts differ significantly from those of the neutral ligand L₁, which anticipate a distinct molecular arrangement in the self‐assembled structure. Although for the neutral ligand the pyridyl protons (H_a and H_b) experience negligible shifts upon cooling, these protons shift upfield for the protonated form (see comparison in Figures 2 a and 4 c). This distinct pattern suggests a parallel stacking of the protonated monomer units within the aggregate, in contrast to the antiparallel arrangement23 observed for the neutral form. To determine the aggregation mechanism and the corresponding thermodynamic parameters of L₁:TFA in MCH, the changes in the chemical shift of H_c were monitored upon temperature variation. The plot of α _agg against temperature shows a sigmoidal curve that can be fitted to the isodesmic model (Figure S4, Supporting Information). The self‐assembly process is enthalpy driven (ΔH=−89.24 kJ mol⁻¹) with an association constant (K) of 1171 m ⁻¹. This calculated K value is approximately twice larger than that of the neutral ligand L₁. The T_m for the self‐assembly process is 324 K (Table S2, Supporting Information). In accordance with UV/Vis absoprtion spectroscopy studies, cooling the L₁:TFA mixture below 303 K in VT NMR led to the formation of higher order aggregates as evident from the broadening and disappearance of NMR signals (Figure 4 c, bottom spectrum), most likely due to the activation of secondary interactions. The rearrangement from an anti‐parallel organization of L₁ to a parallel organization of L₁:TFA in MCH may be caused by the presence of an electrostatic environment created on the pyridyl nitrogen and decreased solubility of the polar protonated species in the nonpolar (MCH) medium.

To examine the morphology of L₁:TFA (1:1) complex in MCH, AFM imaging has been perfomed by spin coating a 5×10⁻⁴  m aggregate solution onto HOPG. Interestingly, micrometer long and entangled helical fibers with 20–40 nm height and 200–400 nm width were observed, as shown in Figure 5. Notably, these fibers do not lead to gelation but rather to the formation of red precipitates when the concentration of the protonated species is increased above 5×10⁻⁴ M^. These results contrast with those previously observed for the neutral ligand L₁, in which a strong gelation ability was found. Thus, it can be concluded that protonation of the pyridyl group of L₁ does not cause the disassembly of the fibers into the free monomers, as commonly observed for various pH‐responsive organogelators,9b, 9e but rather to a new molecular rearrangement leading to distinct aggregate morphology. No circular dichroism (CD) signals could be oberved upon cooling from 363 to 283 K (c=5×10⁻⁴  m; Figure S5, Supporting Information), indicating a non‐preferred (racemic mixture of P and M helices) organization of the achiral L₁:TFA in MCH.

Figure 5.

AFM images of L₁:TFA (1:1; 5×10⁻⁴  m) in MCH prepared by spin‐coating onto HOPG (inset: cross‐section analysis along the yellow line).

The proposed acid‐driven parallel molecular arrangement of the L₁:TFA complex was investigated and optimized using the dispersion‐corrected PM6 semiempirical method (Scheme 1, bottom right; Figure 6 and Figure S6 in the Supporting Information). The calculations suggest that a L₁:TFA dimer (L₁:TFA:TFA:L₁) represents the main structural element of the helical supramolecular structure (see also Movie S1 in the Supporting Information). Electric dipole moments are optimally cancelled inside each dimer, the carbonyls of the amide groups being antiparallel to each other. TFA units are also oriented antiparallel within each dimer. One proton is shared between the pyridine nitrogen and the carboxylic oxygen of TFA, giving rise to an N⋅⋅⋅H⋅⋅⋅O bridge, which enables dimer formation (see Figures 6 and Figure S6 in the Supporting Information). In addition, aromatic interactions and amide H‐bonding between neighboring dimers further stabilize the helical structure, which is in accordance with the VT UV/Vis and VT NMR results. Van der Waals interactions between alkyl chains are also important to stabilize the supramolecule, as shown in Scheme 1.

Figure 6.

Dispersion‐corrected PM6 optimized helix of the L₁:TFA dimer before (top) and after (bottom) UV‐light irradiation, with the highlighted azobenzene groups in the trans (top) or cis (bottom) conformation.

Photoresponsive behavior of L₁:TFA assembly

Irradiation of the L₁:TFA helical aggregates in MCH with a UV LED for 30 min led to the transformation of the initially turbid solution to a clear solution without any color change (Figure 7 a, inset). The absence of any significant hypsochromic shifts demonstrates that the L₁:TFA pairs remain stable upon UV irradiation without disassembly into the free ligand L₁. However, the fact that marked hyperchromic UV/Vis absorption changes are observed upon UV irradiation (Figure 7 a) are in accordance with a possible strain‐driven disassembly.24 Further keeping the solution at 298 K for 3 hours in the dark showed a decrease in absorbance without negligible shifts, indicating a reconstruction of the aggregates. If the solution was kept at 298 K for longer than 3 hours, slow precipitation occurred, possibly due to the formation of higher order assemblies. ¹H NMR measurements were performed to monitor the reconstruction as shown in Figure 7 b. As expected, the initial NMR spectrum of the higher order helical assemblies of L₁:TFA is poorly defined (Figure 7 b, top). However, the appearance of sharp signals of the trans‐azobenzene moieties right after irradiation of the L₁:TFA aggregates for 30 minutes (denoted as t=0) is in accordance with UV changes and suggests a light‐induced disassembly process. This observation can be ascribed to the trans‐to‐cis photoisomerization of the azobenzene and subsequent rapid thermal cis‐to‐trans isomerization.25 Prolonged incubation at 298 K for approximately 12 hours under dark conditions after photoirradiation causes the complete disappearance of the NMR signals, most likely due to the formation of a new assembly that is large in size. AFM measurements on HOPG further supported the reconstruction of assemblies. The images reveal short fibers with a length of approximately 500 nm and height of 1–2 nm, as shown in Figure 7. When the solution was incubated for 12 hours under dark at 298 K followed by spin‐coating on a HOPG surface, only bundling of thin and short fibers were observed (Figure 7 d). Such a transformation of long entangled network of fibers into short and thin fibers after photoirradiation under isothermal conditions indicates that the self‐assembly of L₁:TFA can proceed through a different pathway, as also recently proposed for an azobenzene amphiphile in a liquid crystalline state.25 The absence of any shifts in the UV/Vis spectra of the thin fibers compared to the superhelices (Figure S7, Supporting Information) suggests that the molecular packing is nearly unaffected upon light irradiation and subsequent photo‐reconstruction. Only a weak additional shoulder band around 525 nm is observed for the thin fibers, which might suggest slight differences in translational and/or rotational offsets of the molecules for the two assemblies.

Figure 7.

Time‐dependent a) UV/Vis absorption (pink: 0 min; blue: 180 min after UV irradiation for 30 min at 298 K; inset showing corresponding photographs) and b) ¹H NMR changes of L₁:TFA (1:1, 5×10⁻⁴  m) in MCH‐d ₁₄: before (i) and 0 min (ii), 2 h (iii), and 12 h (iv) after UV irradiation for 30 min at 298 K. c,d) AFM images of L₁:TFA (1:1; 5×10⁻⁴  m) in MCH on HOPG after incubating the irradiated sample at 298 K for 12 h (inset of panel d: cross‐section analysis along the yellow line).

Conclusion

In summary, we have elucidated the effect of acid addition on the self‐assembly pathways of a photoresponsive ligand (L₁). In the absence of acid and light, L₁ self‐assembles in MCH into micrometer‐sized twisted fibers and gels through strong amide H‐bonding and aromatic interactions of antiparallel‐arranged monomer units. Unlike multiple examples of previously reported acid‐sensitive gelators, addition of TFA does not simply lead to a nanostructure disassembly, but rather to the molecular reorganization through a different pathway into superhelical braid‐like fibers. A combination of NMR and UV/Vis absorption spectroscopies as well as AFM and theoretical calculations suggested that L₁ dimers held together by two pyridine–TFA N⋅⋅⋅H⋅⋅⋅O bridges are the main structural elements of the supramolecular helical structures. A preferred parallel stacking of these L₁ dimers supported by amide H‐bonds and van der Waals interactions involving the alkyl chains enables the formation of superstructures, an arrangement that also serves to shield the polar groups from the nonpolar environment. The orientation of the TFA counteranions in an antiparallel fashion further stabilizes the helical assembly by cancelling the electric dipole moments. Ultimately, the photoresponsive behavior of the L₁:TFA species was also examined. Notably, a plausible reconstruction of the initial superhelical fibers into short fibers occurs through a strain‐driven disassembly and subsequent reconstruction. Overall, our findings may contribute to an increased understanding of the relationship between acid and light and their mutual impact on supramolecular polymerization processes.

Experimental Section

Synthetic procedures, materials and methods: The synthesis of L₁ has been previously reported.20 The xerogel was prepared by heating the gel to sol and a few mL of sol was quickly dropped on to HOPG substrate. On leaving at room temperature, the gel was formed and the solvent was evaporated under vacuum before measurement. L₁:TFA dimer was prepared by dissolving equimolar amounts of L₁ and TFA in CHCl₃ and subsequent evaporation of solvent under ambient conditions. The sample was dried under vacuum before using it for further experiments.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

Supplementary

K. K. Kartha, N. K. Allampally, S. Yagai, R. Q. Albuquerque, G. Fernández, Chem. Eur. J. 2019, 25, 9230.