Torsten Dünnebacke,
Kalathil K. Kartha,
Johannes M. Wahl,
Rodrigo Q. Albuquerque
and
Gustavo Fernández
*
Organisch-Chemisches Institut, Westfälische-Wilhelms Universität Münster, Corrensstraße, 40, 48149 Münster, Germany. E-mail: fernandg@uni-muenster.de
First published on 11th September 2020
Control over the photochemical outcome of photochromic molecules in solution represents a major challenge, as photoexcitation often leads to multiple competing photochemical and/or supramolecular pathways resulting in complex product mixtures. Herein, we demonstrate precise and efficient control over the photochemical behaviour of cyanostilbenes in solution using a straightforward solvent-controlled approach based on supramolecular polymerization. To this end, we designed a π-extended cyanostilbene bolaamphiphile that exhibits tuneable solvent-dependent photochemical behaviour. Photoirradiation of the system in a monomeric state (in organic solvents) exclusively leads to a highly reversible and efficient E/Z photoisomerization, whereas a nearly quantitative [2 + 2] photocycloaddition into a single cyclobutane (anti head-to-tail) occurs in aqueous solutions. These results can be rationalized by a highly regular and preorganized antiparallel J-type arrangement of the cyanostilbene units that is driven by aqueous supramolecular polymerization. The presented concept demonstrates a novel approach towards solvent-selective and environmentally friendly photochemical transformations, which is expected to broaden the scope of supramolecular polymerization.
In addition to organic solvents, the use of water as medium for self-assembly is particularly relevant considering its environmentally friendly nature and its relevance in biological processes.15,16 In contrast to organic solvents, self-assembly in aqueous media is relatively less predictable given the difficulty in controlling the competition between hydrophobic and other non-covalent interactions.17,18 However, using rational design strategies, i.e. hydrophilic/hydrophobic ratio, and precisely modulating different experimental parameters, control over aqueous self-assembly processes can be achieved.16,19,20
In this context, programmed supramolecular polymerization represents a versatile tool to finely tune the spatial organization of molecules,1–4,10,21 which may be advantageous for controlling the outcome of photocycloadditions in solution. Expanding this strategy to aqueous media would be particularly relevant,16,17,20,22 as the poor solubility of most organic reactants undergoing photocycloadditions in water requires additional steps of encapsulation and release processes associated with the use of a nanocontainer.23 Given the need for green and sustainable chemical processes,15 supramolecular approaches in aqueous media would help to make photochemical transformations more simple, selective and environmentally benign.
Inspired by these principles, we selected cyanostilbenes as substrates due to their well-known ability to undergo [2 + 2] photocycloadditions and their strong bias towards predefined molecular organization via CN⋯H-bonds.24,25 In addition, cyanostilbenes exhibit excellent photophysical, electronic and stimuli-responsive properties,26,27 which have allowed their implementation in optoelectronics and as biomaterials.28 However, an inherent challenge of cyanostilbenes is the difficulty in controlling their photochemistry, as multiple pathways such as [2 + 2] cycloadditions, photocyclizations7,25,29,30 or E/Z photoisomerizations31 often occur concurrently. In recent elegant examples, microcrystallization methods29 and molecular confinement in nanocontainers32 have been used by Tang and co-workers to achieve regio- and stereoselective [2 + 2] photocycloadditions of cyanostilbenes. Notably, efficient and selective [2 + 2] cycloadditions were achieved for a cyanostilbene-pyridinium salt in aqueous suspensions using UV light.29 By modifying the experimental conditions, the authors could also control the formation of photocyclization and photoisomerization products. Additionally, the photocycloaddition product could be detected by exposure of crystals to room light for two weeks. This relatively long exposure time can be ascribed to the low penetration depth of light in solid matter. Despite these recent advances, “green” supramolecular approaches to control the photochemical outcome of cyanostilbenes in homogeneous aqueous solutions have not been accomplished to date.
Herein, we demonstrate precise and efficient control over E/Z isomerization and [2 + 2] photocycloaddition of cyanostilbenes in solution using visible light via a straightforward solvent-controlled approach based on supramolecular polymerization. This could be achieved by rational design of a linear cyanostilbene bolaamphiphile (Z)-CN-OPE, where the cyanostilbene core is conjugated with two oligophenyleneethynylene (OPE)-based aromatic wedges featuring peripheral glycol chains (Fig. 1a). While the extension of the π-surface is expected to simultaneously enhance the aggregation tendency of the cyanostilbene moiety and shift the irradiation wavelength to the visible region, the incorporation of glycol chains should ensure sufficient solubility in aqueous media. We anticipate that this appropriate balance of hydrophilic and hydrophobic structural elements would enable the required topochemical control for a [2 + 2] photocycloaddition to occur efficiently and selectively in aqueous media without the need of additional templating or confinement processes. On the other hand, the absence of close intermolecular contacts in organic “good” solvents is expected to hinder the photocycloaddition and preferentially favor a reversible E/Z photoisomerization pathway. In order to understand all these photochemical process in detail, both geometrical isomers of the target molecule (Z)- and (E)-CN-OPE were synthesized separately and fully characterized (see ESI for details†).
On the other hand, the (E)-CN-OPE isomer showed no significant differences in the UV-Vis spectra (c = 10−5 M) (Fig. 1d and S14†) upon changing the solvents. Only in water, the occurrence of hypochromism is indicative of aggregation. However, the absence of additional spectral changes, i.e. shifts, suggests a weaker and less ordered aggregation than the linear (Z)-CN-OPE isomer. This assumption can be further supported by emission lifetime measurements (Fig. S15†). Despite that the 12-fold increase in the average emission lifetime of (E)-CN-OPE in aqueous solutions (0.141 ns in THF vs. 1.674 ns in H2O) implies an occurring aggregation process, the resulting average lifetime for aggregated (Z)-CN-OPE is significantly higher (4.471 ns in H2O), indicating a less ordered structure for (E)-CN-OPE in aqueous solutions. Furthermore, all attempts to obtain AFM images of these solutions showed no identifiable objects (Fig. S16†).
The photoluminescence (PL) spectra of both isomers are comparable in most of the tested solvents, with emission maxima between 455 and 472 nm (Fig. 1c, e and S17†), indicating that both isomers have the same emissive state.33,34 This is further validated by nearly identical average lifetimes of both isomers in THF (Fig. S15a and b†). The comparatively larger Stokes shifts (ΔλST) observed for (E)-CN-OPE (ΔλST = 109–128 nm) compared to (Z)-CN-OPE (ΔλST = 63–82 nm) (Fig. 1c, e and Table S1†) are also in agreement with different ground states, regardless of the same emissive state. Significantly different PL spectra were found for these isomers in water (Fig. 1c, e and S17†), which can be due to the effect of aggregation. For both isomers, the substantial bathochromic shift in the emission band (198 nm for (Z)-CN-OPE and 172 nm for (E)-CN-OPE) corroborates a J-type exciton coupling, which is more pronounced for the (Z)-CN-OPE isomer (Fig. 1c, e, S17 and Table S1.† For further details on the photophysical behaviour of both isomers, see ESI†).
Due to the strong aggregation propensity of linear (Z)-CN-OPE in water at room temperature even at concentrations as low as 3 × 10−7 M, the use of a co-solvent was necessary for mechanistic investigations.35 This approach also helps attenuate the marked lower critical solution temperature (LCST) behaviour of (Z)-CN-OPE in pure water (Fig. S18†), by which dehydration of the glycol chains upon heating (for the present system above 328 K at 10−5 M) results in turbid dispersions (see ESI for details†).36 Addition of a co-solvent (THF, volume fraction of water (fw) < 0.80, Fig. S18d†) or decreasing the concentration to 6 × 10−6 M in combination with heating enabled the derivation of the thermodynamic parameters of the self-assembly. Absorption and emission studies of (E)- and (Z)-CN-OPE (c = 10−5 M) in THF/water mixtures were initially carried out (Fig. 2a and S19†). Addition of up to 70 vol% water (fw ≤ 0.70) leads to an overall increase in absorption of (Z)-CN-OPE without any shift in λmax due to a possible restriction of internal rotation around the Φ1 to Φ6 bonds and associated planarization of the OPE scaffolds (see Fig. 1a).37 A comparison of the fluorescence excitation spectra in THF and water suggests an aggregation-induced planarization of the molecules (Fig. S20†). While the existence of two distinct excitation spectra in THF indicates different conformations of (Z)-CN-OPE in the molecularly dissolved state, a more planarized species is preferred in aqueous media. Upon further addition of water (fw > 0.75), a bathochromic shift (Δλmax = 18 nm), accompanied by a shoulder formation around 450 nm and a decrease in absorption are observed. This red-shifted absorption band further supports the formation of J-type aggregates.38 Several cooling rates (0.1 K min−1 to 5 K min−1) and THF/water ratios (fw = 0.70–0.80) were applied to demonstrate that (Z)-CN-OPE only forms a single, thermodynamically favored aggregate in solution (Fig. S21a–e†). The plots of αaggvs. temperature for (Z)-CN-OPE in water yielded non-sigmoidal curves (Fig. 2d) that were fitted to the cooperative nucleation–elongation model (Fig. S21a–e†),39 as also observed for other OPE-based bolaamphiphiles.40 In contrast to (Z)-CN-OPE, UV-Vis absorption studies of the V-shaped isomer (E)-CN-OPE at different THF/water ratios (fw = 0.10–1.00) showed no significant shifts in the λmax apart from a slight hypochromic behaviour (Fig. S19†).
Subsequently, fluorescence changes of (Z)-CN-OPE were monitored in different THF/water mixtures (fw = 0.10–1.00) as shown in Fig. 2b. Initial addition of 10% vol water (fw = 0.10) to a THF solution of (Z)-CN-OPE results in a drastic reduction of the fluorescence with no shifts in the emission maxima (λmax = 487 nm, Fig. 2b). Further addition of water induces a gradual red-shift in the emission band, which reaches a maximum of 570 nm at volume fractions of 80% (fw = 0.80). Simultaneously, the fluorescence is enhanced by 1.5 fold compared to that in pure THF (Fig. 2b). This behaviour can be attributed to the aggregation-induced emission enhancement (AIEE) commonly observed for linear (Z)-cyanostilbenes,41,42 which is caused by the restriction of internal rotation around the CC axis upon photoirradiation.33,34 Increasing fw beyond 0.80 shows no significant changes in the fluorescence, indicating saturation of aggregation. On the other hand, the main emission band of (E)-CN-OPE undergoes a less significant red-shift in solvent-dependent experiments, even using large water contents (fw > 0.85; see Fig. S19d†). Thus, based on absorption and emission properties, only the more preorganized, undistorted linear isomer (Z)-CN-OPE is prone to efficient aggregation.
To elucidate the molecular packing, 1H NMR experiments were recorded by adding increasing volume fractions of D2O (fD = 0.0 to 0.3) to a monomer solution of (Z)-CN-OPE in THF-d8 (Fig. 2c, S22 and S23†). Addition of up to 10% vol water (fD = 0.0–0.10) causes a weak upfield shift of the aromatic signals (Δδ = 0.03 ppm) along with a more significant downfield shift for the signal of the vinylic proton (Δδ = 0.10 ppm) (Fig. 2c). Further addition of D2O (fD > 0.1) leads to an overall upfield shift for the signals corresponding to all aromatic rings as well as that of the vinylic proton. This trend, which has been recently observed for hydrophobic linear OPEs,43 points towards a plausible two-step supramolecular polymerization process: (i) an initial pre-nucleation event at fD < 0.10 driven by partial π–π stacking of the OPE units and H-bonding between the vinyl proton of one molecule and the cyano nitrogen of a neighbouring one (Cvinyl–H⋯NC, Fig. 3a, top); (ii) a subsequent elongation process mostly triggered by aromatic and hydrophobic interactions when fD exceeds 0.10. Such a two-step mechanism is also supported by UV-Vis experiments in THF/water mixtures as discussed above.
2D ROESY NMR spectra were recorded at 299 K in THF-d8/D2O (1:
1) to shed more light on the packing of (Z)-CN-OPE. The appearance of a correlation signal between the vinylic proton and the adjacent aromatic protons, which is absent in COSY, corroborates a slight offset-stacking of the monomers of (Z)-CN-OPE (Fig. S24†).
FTIR measurements of (Z)-CN-OPE in THF (monomer) and thin film of an aggregate solution confirmed the existence of intermolecular CN⋯H interactions in the aggregated state, as the CN stretch shifts to lower wavenumbers (from 2216 to 2203 cm−1) upon aggregation (Fig. S25†).
Morphological and topological features of the (Z)-CN-OPE aggregates were obtained from Dynamic Light Scattering (DLS) experiments and Atomic Force Microscopy (AFM) (see Fig. S26 and S27†). Angular-dependent DLS of (Z)-CN-OPE in water (c = 10−5 M) revealed size distributions with average hydrodynamic radii RH of around 67 nm. Although the derived size distributions slightly vary (∼73 nm to ∼60 nm) within this set of measurements, the aggregates can be considered of spherical shape, as the RH values are stable over a wide range of angles (Fig. S26†).44 AFM imaging of the same aggregate solution drop-casted onto mica reveals the formation of disk-like aggregates with an average diameter of ∼80 nm and an average height of ∼3.5 nm (Fig. 2e, S27a and b†). The difference in the observed aggregate size when comparing the results from DLS and AFM can be explained by the hydration shell present in solution. Additional AFM studies increasing the amount of drop-casted material in combination with extended deposition time show the growth of several anisotropic nanostructures (Fig. S27c†). The height profile shows values of 3.5 ± 0.5 nm, which are in the same range as the isolated disks. The obtained height profiles are slightly lower than the length of the aromatic system of (Z)-CN-OPE (ca. 4.2 nm). This suggests that the monomer units within the assemblies are not arranged exactly perpendicular to the mica surface, but rather slightly tilted. Combining these results, we envisaged a stepwise self-assembly process involving a pre-nucleation, nucleation and elongation events that ultimately result in discoidal nanostructures, as illustrated in Fig. 3a.
Quantum mechanical calculations at the semiempirical dispersion-corrected PM6 level were carried out to provide insights on the structural features of self-assembled (Z)-CN-OPE. As shown in Fig. 3a and b, the cyano groups tend to arrange alternately in opposite directions within the optimized tetramer stacks, thereby cancelling the total electric dipole moment.45 The OPE units are planar and, consequently, involved in intermolecular aromatic interactions, whereas the CC double bond displays a distinct out of plane twist. While in isolated tetramer stacks intermolecular CN⋯H hydrogen bonds are not feasible (Fig. 3a, right and S28†), this can efficiently occur if two tetramers interact laterally to form an octamer, as revealed from the optimized stack shown in Fig. 3b. Hereby, the cyanostilbene strands formed by the respective tetramers align in a parallel fashion with a slight offset, allowing two corresponding, non-aromatic substituents attached to the C
C double bond to form a head-to-tail like arrangement (see grey box in Fig. 3b).
The intermolecular distances from the cyano nitrogen to its neighbouring protons are d = 2.5 Å (CN⋯H–CC) and d = 2.3–3.0 Å (CN⋯Harom) (Fig. 3b, grey box). These theoretical results are in good agreement with intermolecular interactions determined by crystallographic data of previously reported cyanostilbenes.29,45,46 Furthermore, the calculated center-to-center distance between two adjacent stilbenic C
C double bonds is 3.7 Å, which is very close to the values found for various stilbene13,47 and cyanostilbene29,48 assemblies undergoing [2 + 2] cycloaddition reactions. On this basis, we envisaged that the ideal dye arrangement of (Z)-CN-OPE can facilitate an efficient and selective light-induced cycloaddition reaction in the self-assembled state and a reversible geometrical photoisomerization in the isotropic state.17,18
Further investigations of (Z)-CN-OPE under more dilute conditions (c = 10−5 M) using UV-Vis spectroscopy resulted in a photostationary state of ≥90% of (E)-CN-OPE. As this concentration range is too low for 1H NMR measurements, a spectroscopic method to determine the E:
Z ratio was also employed (see ESI for details†).
Initially, UV-Vis absorption changes of (Z)-CN-OPE in DCM by adding increasing aliquot amounts of (E)-CN-OPE were monitored to identify the spectral features at different ratios (Fig. 4b). The plot of molar absorptivity ε vs. the percentage (%) of (Z)-CN-OPE yielded a linear regression (Fig. 4b inset). This calibration function was subsequently used to determine the Z:
E equilibrium under light irradiation. Exposing solutions of either (Z)- or (E)-CN-OPE to visible light (λexc = 465 nm) for defined time intervals displays intermediate absorption spectra that cross via a defined isosbestic point.
This supports the presence of only two species (E- and Z-isomers) in equilibrium (Fig. 4c). Plotting the Z:
E ratio as a function of time reveals a plateau after 20 s regardless of the initial pure isomeric ratio (Fig. 4d). The achieved photoequilibrium state mainly consists of the (E)-CN-OPE isomer (6
:
94 (Z
:
E), Fig. 4; for detailed information on the calculated equilibria ratios see ESI†). UV irradiation (λexc = 365 nm) triggers the back isomerization (E-to-Z) by re-establishing the photoequilibrium of 67
:
33% (Z
:
E) (Fig. 4d and Table 1). The time required to reach the equilibrium upon irradiation with UV light was around 2–3 s (Fig. 4d), which is considerably faster than in the presence of visible light. This interconversion between (E) and (Z) is highly reversible, no degradation being observed even after 20 consecutive cycles of alternated irradiation at 465 and 365 nm (Fig. 4e).
Solvent in decreasing order of polarity45 | Equilibrium λirr = 465 nm, Z![]() ![]() |
k Z→E,465 nm/10−2 s−1 | Equilibrium λirr = 365 nm, Z![]() ![]() |
---|---|---|---|
MeOH | 21![]() ![]() |
2.49 | 66![]() ![]() |
CH3CN | 57![]() ![]() |
1.03 | 87![]() ![]() |
DMSO | 57![]() ![]() |
3.08 | 90![]() ![]() |
DCM | 6![]() ![]() |
15.48 | 67![]() ![]() |
THF | 28![]() ![]() |
7.30 | 85![]() ![]() |
Dioxane | 3![]() ![]() |
18.91 | 47![]() ![]() |
Toluene | 6![]() ![]() |
34.01 | 45![]() ![]() |
We also analysed the impact of the solvent on the kinetics and efficiency of the photoisomerization process (Fig. S29 and S30†). The results reveal that the equilibrium concentration of (E)-CN-OPE, as well as the first-order rate constant for the Z-to-E conversion, are dependent on the chosen solvent (Table 1). In rough approximation, the rate constant kZ→E is higher in non-polar solvents compared to polar solvents, while the equilibrium E/Z ratio exhibits only a weak trend in solvent polarity. Analysis of the ground and excited states using a Lippert–Mataga-plot50 revealed a more polar excited state (Fig. S31†), as previously reported for polar cyanostilbenes.26,42 We therefore assume that the influence of the solvent on the excited state may affect the reaction rate by possibly facilitating competing relaxation pathways such as twisted intramolecular charge transfer (TICT).51,52 Nevertheless, due to the polar, bulky TEG chains and the conjugated aromatic core, which induce different solvent–solute interactions, other solvent properties like viscosity/solvent friction should also be taken into account.52
Once the specific isomer was identified, we further examined the efficiency of the [2 + 2] photoreaction in water by selecting additional light sources with different energies (λexc = 365 and 520 nm) (Fig. S35a–d†). Irradiation with UV light (λexc = 365 nm) causes an increase in the reaction rate, however at the expense of a slight decrease in the absorbance intensity, which might arise from other, less favourable side reactions (Fig. S35a†). Interestingly, the selective [2 + 2] photocycloaddition also occurs efficiently without side products when the aqueous solution of aggregated (Z)-CN-OPE is exposed to green light (λexc = 520 nm; Fig. S35c†). Plotting the absorption changes at 399 nm vs. time revealed a second order kinetics (Fig. S35a–c†). As the sun remains the most economical and sustainable light source, we finally exposed the aqueous assembly of (Z)-CN-OPE to normal sunlight and followed the photochemical reaction. Nearly complete conversion of (Z)-CN-OPE was observed in only 120 s, indicating the fast and efficient photochemical reaction under green conditions (normal daylight and in water; Fig. S35d†).
Finally, to further demonstrate the advantage of a supramolecular approach in aqueous media, we synthesized a structurally related (Z)-CN-OPE* (see Fig. 1), in which the aromatic surface is reduced at each side of the stilbene moiety (for synthesis and characterization, see ESI†). Even though the aggregation tendency is reduced by this structural modification (Fig. S36a†), we were able to reproduce the solvent-dependent photochemical behaviour observed by (Z)-CN-OPE. While irradiation of (Z)-CN-OPE* in the molecularly dissolved state with either visible or UV light (λ = 465 nm and 365 nm) triggers the reversible E/Z isomerization (Fig. S36c and d†), an efficient (90% yield) photocycloaddition into a single product (cyc-CN-OPE*) occurs in water (Fig. S36b and S11†), in good agreement with the behaviour of (Z)-CN-OPE.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d0sc03442h |
This journal is © The Royal Society of Chemistry 2020 |