Design and Synthesis

We first explored the supramolecular monomer S-BOR, with para-benzamide units on the boroxine ring (Figure 1b). Molecular modelling of S-BOR indicated supramolecular polymerization into one-dimensional helical fibres via threefold hydrogen-bonding of the peripheral amides and π–π stacking of the aromatic benzene units (Figure S1). We further introduced homochiral solubilizing side chains to bias the helicity of the supramolecular polymer and thereby enable assembly studies with circular dichroism (CD) spectroscopy.

The respective boronic acid precursor S-BAc was synthesized in one step from the commercially available 4-boronobenzoic acid and (S)-3,7-dimethyloctan-1-amine under Steglich amidation conditions in 80 % yield. Upon self-condensation of S-BAc, the boroxine can be obtained after the elimination of three equivalents of water. This process is entropically favoured, and the equilibrium is shifted towards the boroxine at elevated temperatures. Following the principle of Le Chatelier, the removal of water further shifts the equilibrium towards the boroxine.

We investigated two different strategies to synthesize S-BOR from the boronic acid S-BAc, in bulk as well as in solution. Heating the bulk boronic acid in a vacuum oven at 110 °C overnight resulted in the quantitative conversion to the boroxine as evidenced by the ¹H NMR displayed in Figure 2a (see Figure S2 for full spectrum). The disappearing boronic acid signal at 8.17 ppm as well as the slight downfield shift of the aromatic protons from 7.77 to 7.82 ppm and 7.84 to 7.93 ppm indicates boroxine formation. Additional DOSY experiments showed an increase of the hydrodynamic radius (R_H) by a factor ~1.8 (Figure S3), which matches the expected increase in size upon trimerization. Further, in agreement with FT-IR spectroscopy of boroxine compounds,²⁷ the characteristic boronic acid vibrations at 1115, 1006 and 803 cm⁻¹ disappeared upon condensation and the diagnostic boroxine vibrations at 705 and 678 cm⁻¹ appeared (Figure 2c). Using variable temperature IR, the sudden increase in intensity of the boroxine vibration at 705 cm⁻¹ above 80 °C indicates that conversion to boroxine occurred around this temperature (Figure 2d, full spectra shown in Figure S4). In contrast to the work of Aida and co-workers on solvent-free autocatalytic supramolecular polymerizations on phthalocyanines,⁴⁶ we could not identify any autocatalytic properties in our system.

The boroxine formation is further supported by differential scanning calorimetry (DSC) characterization, where the DSC pan was not hermetically sealed to allow the forming water to evaporate (Figure 2e). The large endothermic transition visible from 75 °C onwards can be attributed to the boroxine formation. Upon heating under closed conditions, the pressure build-up in the DSC pan shifted the endothermic transition of the boroxine formation to 141 °C (Figure S5). Under the polarized optical microscope (POM), boronic acid S-BAc showed no birefringent domains, indicating the absence of ordered structures prior to trimerization (Figure 2b). Upon driving self-condensation by heating the sample to 100 °C under a constant nitrogen flow, birefringent domains became evident, stipulating the presence of ordered microstructures. We reason that upon formation of the C₃-symmetrical boroxine, supramolecular assembly becomes feasible via hydrogen-bonding and π–π stacking, causing the occurrence of birefringent domains. This is further supported by VT-IR measurements, where a shift from 1635 to 1632 cm⁻¹ of the amide I vibration indicates supramolecular aggregation via hydrogen-bonding (Figure S6). The high dynamicity of the boronic acid-boroxine equilibrium resulted in fast hydrolysis (within seconds) under ambient conditions, thus making further sample preparation challenging.

We therefore aimed to optimize the boroxine synthesis in solution, where the boroxine hydrolysis can be limited by using drying reagents. Initial solution trimerizations were attempted in polar protic DMSO-d6 (30 mM) over 3 Å molsieves and were monitored by ¹H NMR. No conversion from S-Bac to S-BOR could be observed under these conditions, even at elevated temperatures (Figure S7). We reason that the trimerization in hydrogen-bonding solvents might be energetically less favourable as the boronic acid and the open forms, dimer and trimer, are stabilized by strong hydrogen-bonding with the respective solvent. DFT calculations of the trimerization in DMSO and apolar cyclohexane confirm this hypothesis as the trimer formation appears to be less favoured in polar protic solvents (Figure 3a).

We thus changed the solvent system to apolar decalin-d18. Upon stirring S-Bac in decalin-d18 for 2 h at room temperature over 4 Å molsieves, the complete dissolution of S-Bac indicated conversion to the less polar boroxine, but aggregation resulted in the absence of any traceable ¹H NMR peak (Figure S8). Therefore, we switched to the polar aprotic solvent TCE-d2. Upon dispersing S-Bac in TCE-d2 (30 mM) over 4 Å molsieves for 2 min, the ¹H NMR spectrum showed the presence of S-Bac and S-BOR (Figure 3b). After leaving the sample for 5 h at room temperature, full conversion was obtained. Similar to the trimerization in bulk, DOSY showed an increase in R_H from 0.55 nm to 0.93 nm, stipulating S-Bac trimerization in solution, which was further confirmed using mass spectrometry (Figures S9–S10). Under inert conditions, no degradation of the boronic acid could be observed over months (Figure S11), prompting us to perform further sample preparation in the glovebox.

Assembly of S-BOR Into 1D Supramolecular Polymers

To study the aggregation of S-BOR, the concentrated stock solution in TCE-d2 (10 mM) was diluted with decalin (DEC, 25–100 μM), equilibrated for 5 min and monitored by CD and UV/Vis spectroscopy (Figure 4). In 1 vol % TCE-DEC, a positive bisignate Cotton effect with a maximum at 257 nm was observed (Figure 4b), suggesting the formation of chiral supramolecular aggregates. The zero crossing of the CD signal coincides with the absorption maximum, indicating exciton coupled chromophores in the aggregate (Figure S12).

To assess the stability of the supramolecular polymers, we then performed denaturation experiments by preparing multiple samples with increasing amount of TCE while keeping the concentration constant (50 μM), following the sample preparation protocol described above. Increasing the amount of TCE resulted in an equilibrium shift from supramolecular polymer to monomer, as visible in UV/Vis and CD spectroscopy (Figure 4b and c). Upon TCE addition, the absorbance intensity increased and a slight red shift from 245 nm to 247 nm was observed, indicative for the disassembly of H-aggregates (Figure 4c). A thermodynamic mass-balance model was fitted to these denaturation experiments at three individual concentrations (25, 50 and 100 μM, Figure 4d) to determine the thermodynamic parameters of the supramolecular polymerization. The Gibbs free energy change for elongation was ΔG_e=−35.4 kJ mol⁻¹ with a nucleation penalty (NP) of 12.9 kJ mol⁻¹, suggesting a cooperative supramolecular polymerization mechanism.⁴⁷ Further, the linear dependency of ΔG_e on the fraction of TCE is described by m_TCE and was determined as 47.7 kJ mol⁻¹ (Section 7 in Supporting Information).

We further attempted to study temperature-induced disassembly. During the first heating run, we observed the expected decrease in CD intensity upon thermal dissociation of the supramolecular polymer (Figure S13c). However, the subsequent cooling run did not show a complete recovery of the initial CD intensity, stipulating possible decomposition of the boroxine. Interestingly, the absorbance showed a slight blue shift after the initial heating-cooling run, suggesting the formation of a less conjugated system, like the boronic acid (Figure S13a). In addition, the used cuvettes appeared to show a stain after removal of the solution, which could not be easily cleaned with common organic solvents (Figure S13d). This indicates that the free hydroxy groups of the quartz-glass might react with boroxine to form boronic ester-functionalized quartz-glass, impeding variable temperature measurements. At room temperature however, the CD intensity remained constant over the course of multiple hours when the samples were kept over molsieves (c=100 μM, 5 vol % TCE/DEC, Figure 4e), pointing towards high boroxine stability in conditions where water is constantly removed from the system. In the absence of molsieves, disappearance of the CD signal was observed over the course of 17 h alongside precipitation. This suggests that the few ppm (≈10 ppm) of water remaining in the pre-dried solvents is enough to hydrolyze the boroxine and cause disassembly. Considering a water saturation of 10 ppm in pre-dried decalin and a sample concentration of 100 μM, the ratio of water to boroxine is ≈1. Thus, even when working under dry conditions, water is approximately equimolar to boroxine, which explains the observed hydrolysis.

Copolymerization Studies and Amplification of Asymmetry

In the past, mixing of chiral and achiral monomers in sergeant-and-soldiers experiments has significantly contributed to the understanding of the amplification of asymmetry in supramolecular co-polymerizations.⁵⁰ Without a stereocentre in the monomer, discotic supramolecular monomers typically polymerize into equal amounts of left- and right-handed helices with no detectable CD signal, as shown by the achiral derivative n-BOR (Figure S14). We were intrigued to investigate the amplification of asymmetry in boroxine-based supramolecular polymers and compare mixtures of chiral and achiral boronic acid precursors with chiral and achiral boroxine trimers. In the in situ sergeant-and-soldiers experiment, both sergeants and soldiers were prepared in situ by co-condensation of chiral S-Bac with achiral n-Bac (Figure 5a), while in the classical approach, chiral S-BOR and achiral n-BOR monomers were co-polymerized. First, the chiral and achiral boronic acids were mixed in ratios ranging from 0 to 100 mol % of S-Bac to n-Bac in TCE-d2 (30 mM) and were co-reacted as previously described for the homo-condensation. We employed MALDI mass spectrometry to analyse these mixtures by comparing the intensity of the [M+Na]⁺ peaks, assuming that all four derivatives have similar ionizability (Figure S15b–g). Four different hetero-boroxine monomers in a 1 : 3 : 3 : 1 distribution of S-BOR : S¹-n²-BOR:S²-n¹-BOR : n-BOR, respectively, were obtained from a 1 : 1 mixture of S-Bac and n-Bac (Figure S15a,g). The statistical distribution indicates similar reactivities for the chiral and achiral boronic acids (Table S1). ¹H NMR analysis did not show any traceable difference between the different species (Figure S16), limiting our analysis to MALDI mass spectrometry.

We diluted the obtained TCE-d2 stock solutions of the co-condensates with decalin to trigger assembly. Amplification of asymmetry was observed in the mixed samples, with about 20 mol % of chiral S-Bac necessary to reach full helical bias (Figure 5b, beige circles). MALDI experiments indicated that n-BOR and S¹-n²-BOR are the major fractions at mixing ratios below 20 mol % of S-Bac (Figure S15b–f). This stipulates that the amplification of asymmetry in the system depends solely on the presence and quantity of point chirality in the side chains, irrespective of the composition of the boroxine monomer.⁵¹ We further performed the classical sergeant-and-soldiers experiment by mixing chiral S-BOR with achiral n-BOR. The amplification of asymmetry seemed to be nearly identical (Figure 5b, blue circles), prompting us to investigate the chemical composition after mixing of S-BOR and n-BOR. MALDI experiments show the presence of the hetero-boroxines within seconds after sample preparation (Figure S17). This suggests fast intermolecular boronic acid exchange between the boroxines,²⁸ yielding similar statistical mixtures of hetero-boroxines as obtained after in situ co-condensation and formation of supramolecular polymers of variable composition.

Co-Condensation with Pyrene Boronic Acid for Circularly Polarized Luminescence (CPL)

The ease of boroxine formation and fast exchange kinetics between boronic acids and boroxines described in the previous section prompted us to elucidate the possibility to co-condense functional boronic acids with S-Bac. We anticipated that co-condensation of S-Bac with functional boronic acids might allow to construct hetero-boroxines that are still able to assemble into ordered aggregates by intermolecular hydrogen-bonding of S-Bac. This would open the possibility to use S-Bac as supramolecular scaffold to arrange functional molecules in space.²¹ The large number of commercially available boronic acid derivatives bearing functional units allowed us to freely choose a co-reactant with the desired properties. Pyrene is well known to exhibit a strong excimer emission upon aggregation, which provides a useful readout of potential aggregation using fluorescence spectroscopy (Figure 6a).⁵² Thus, as a proof of concept, we studied the co-condensation of pyrene boronic acid (Py-Bac) with S-Bac. Upon mixing of S-Bac and Py-Bac in a 5 : 1 ratio in 10 mM TCE-d2 solution, three different boroxine species were observed in ¹H NMR spectroscopy (Figure S18a), that were identified as S-BOR (53 %), S²-Py¹-BOR (37 %) and S¹-Py²-BOR (10 %). The homo-condensate Py-BOR was not detectable by ¹H NMR and only traces appeared in MALDI experiments (Figure S18b). Isolating the different co-condensates was impeded by the high hydrolysis sensitivity of the boroxine. We thus decided to use the mixture as obtained to conduct studies into the chiroptical properties of the system.

After dilution of the 10 mM TCE-d2 stock solution with decalin (50 μM, 5 vol % TCE/DEC), a positive bisignate CD signal was observed similar in shape to the CD signal of S-BOR homopolymers (Figure 6b). Additionally, the presence of a weak CD signal between 340 nm and 400 nm, coinciding with the new absorption feature arising from pyrene (Figure 6d), indicated the chiral assembly of pyrene in solution. This is surprising, as the presence of the pyrene disrupts hydrogen-bonding in the supramolecular polymer. This indicates that the mixed Py-Bac/S-Bac monomers are still able to form ordered supramolecular aggregates despite the reduced amounts of stabilizing intermolecular hydrogen bonds. Moreover, fluorescence measurements at 10 mM concentrations showed strong pyrene-excimer emission around 460 nm with only negligible contributions from pyrene-monomer emission around 410 nm (Figure 6c). The high concentration resulted in gelation of the mixture, as shown in the inset of Figure 6c. In pure TCE (10 mM), only weak excimer emission could be detected, strongly suggesting that the excimer emission obtained at 10 mM in 5 vol % TCE/DEC is intermolecular in nature, and does not arise from intramolecular excimer formation in Py-BOR or S¹-Py²-BOR monomers. At lower concentrations (50 μM) in 5 vol % TCE/DEC, the pyrene-monomer emission is more pronounced, but the pyrene-excimer emission remains visible, confirming the presence of pyrene-pyrene interactions in the supramolecular polymer, even at low concentrations (Figure S19). Additional CPL measurements of the mixture in 10 mM decalin showed a dissymmetry in emission in the order of g_PL ~9×10⁻⁴ (Figure S20), stipulating that the pyrene is aggregated in a chiral fashion⁵³ as also indicated by the CD response at lower concentrations. The presence of the intermolecular and circularly polarized pyrene-excimer emission stipulates stacked pyrene-pyrene moieties in a chiral orientation, indicating possible incorporation of pyrene into the 1D supramolecular polymers of S-BOR. Unfortunately, we were unable to determine a precise molecular picture of the aggregate microstructure, but the chiroptical results confirm the use of boroxine as supramolecular scaffold for CPL-active supramolecular polymers.