Formation of pseudopolyrotaxanes. First, we envisioned using non-alkylated pillar[5]arene P5OH46 as the ring components of PEO-based pseudopolyrotaxanes (Fig. 2a). Because of the 10 phenolic OH groups, P5OH is able to form effective hydrogen bonds and is partially soluble in aqueous media, unlike alkylated pillar[n]arene derivatives. In this study, P5OH was dissolved in a mixture of water and methanol [1/1 (v/v)], to which was added a solution of OH-terminated PEO (molecular weight: 2000) PEO2k-OH in the same solvent system. Threading of PEO2k-OH into P5OH yielded the pseudopolyrotaxane P5OH•PEO2k-OH as a precipitate owing to its decreased solubility in the mixed solvent (Supplementary Movie 1). The obtained solids were washed with water to remove free P5OH and PEO2k-OH and slowly dried by heating at 60 °C in air.
Pseudopolyrotaxane P5OH•PEO2k-OH was dissolved in acetone-d6 and characterized by 1H NMR spectroscopy (Supplementary Fig. 2). Sharp peaks from dissociated P5OH and PEO2k-OH appeared at 7.96, 6.67, and 3.59 ppm as a consequence of de-threading.46 On the basis of the integral values, the molar ratio of ethylene oxide (EO) units to P5OH was estimated to be 8. A solid sample of P5OH•PEO2k-OH was also investigated by solid-state 2D 1H/13C hetero-correlated NMR spectroscopy with magic angle spinning (Fig. 3a). Cross-peaks were detected between the 1H signals due to P5OH and the 13C signal due to PEO2k-OH at 69.7 ppm. These intermolecular correlations clearly indicated the proximity of the two components. Furthermore, the observed 13C peak was shifted to higher field than the peaks of the crystalline and non-crystalline PEOs (71.2 and 72.0 ppm, respectively).47,48 This peak shift could be rationalized in terms of aromatic shielding, thereby indicating that complexation between the PEO chains and pillar[5]arene rings had occurred.22
Powder X-ray diffraction (PXRD) analysis was performed for P5OH•PEO2k-OH and its components, P5OH and PEO2k-OH, to investigate the solid-state structures (Fig. 3b). While the solid sample of each component showed several sharp peaks in the PXRD pattern, the pattern for the pseudopolyrotaxane P5OH•PEO2k-OH was composed of unclear signals. These results indicated the absence of uncomplexed P5OH and PEO2k-OH and the formation of a non-crystalline complex. The absence of free P5OH in the pseudopolyrotaxane sample was also confirmed by the lack of CO2 gas adsorption behavior (Supplementary Fig. 18).
Threading into P5OH changed the assembled structures of the PEOs to cause a large perturbation in the thermal properties. In differential scanning calorimetry (DSC) studies, P5OH•PEO2k-OH showed no explicit signals from − 70 °C to 100 °C, while PEO2k-OH had endothermic and exothermic peaks at 53 and 25 °C, because of the melting and crystallization of the pristine PEO (Supplementary Fig. 20). The disappearance of the melting behavior of PEO was highly dependent on the number of threaded ring units. The molar ratio of EO units to P5OH could be rationally controlled by changing the amount of added PEO solution in the preparation (Supplementary Table 2), providing pseudopolyrotaxanes with EO-to-P5OH ratios of 8–23. As the number of EO units increased from 8 to 15, endothermic peaks gradually became evident and shifted to lower temperatures (Supplementary Fig. 21). Thermogravimetric analysis (TGA) in air was also performed for P5OH•PEO2k-OH along with its uncomplexed components (Supplementary Fig. 23). The 10% weight loss temperatures were 164, 229, and 302 °C for P5OH, PEO2k-OH, and P5OH•PEO2k-OH, respectively, clearly indicating an improvement in the thermal stability upon complexation.
To obtain further insight into the pseudopolyrotaxane formation, we prepared P5OH•PEO2k-OH in the presence of adiponitrile as a competitive guest. The yield of pseudopolyrotaxanes decreased from 32–3.2% and the molar ratio of EO units to P5OH was increased from 8 to 27 as a result of ineffective threading. Pseudopolyrotaxane formation was also hampered when the pH of the solvent system was increased to 12 using NaOH (yield: 8.8%; EO-to-P5OH ratio: 193). Such basic conditions cause deprotonation of P5OH to generate anionic species, which experience charge repulsion with one another. Additionally, we tested the preparation of the pseudopolyrotaxane using pure methanol instead of the mixed solvent but could not obtain any precipitate. Hence, hydrophobic–hydrophilic interactions between the hydrophobic ethylene groups of the PEOs and pillar[5]arenes are thought to be a major driving force to form the pseudopolyrotaxane structure, in addition to the multiple hydrogen bonds between the pillar[5]arene rings. Alteration of the terminal groups of the PEOs did not appreciably affect the threading into the P5OH rings, and amine-, methoxy-, and tosylate-terminated pseudopolyrotaxanes were able to be constructed (Fig. 2a). Further, PEO2k-OH could be replaced with PEOs with larger molecular weights up to 500,000 (Supplementary Table 3). As the PEO molecular weight increased, the yield of pseudopolyrotaxane and EO-to-P5OH ratio tended to increase probably because of decreased solubility in the mixed solvent.
Synthesis of polyrotaxanes. Encouraged by the versatility in PEO axles, we next attempted the synthesis of polyrotaxanes composed of P5OH and PEOs (Fig. 2b). Following the procedure described above, pseudopolyrotaxane P5OH•PEO20k-COOH with an EO-to-P5OH ratio of 4.2 was prepared in 19% yield, and then the terminal carboxylic acid groups were capped by condensation with 1-adamantanamine in acetone, with the assistance of BOP reagent and EDIPA.49 Before the completion of the amide formation, some ring units were de-threaded and the resulting free P5OH and PEO20k-COOH were dissolved in acetone. In contrast, the capped polyrotaxane P5OH•PEO20k-CONHAd was not very soluble in acetone, enabling purification by simple washing with acetone. Intermolecular hydrogen bonds involving P5OH would prevent the solvation in acetone. The polyrotaxane was soluble in a limited number of solvents, including DMF, DMSO, and basic water. Similar results have been also reported for polyrotaxanes constructed from α-CD and PEO.19
The 1H NMR spectrum of P5OH•PEO20k-CONHAd was recorded in DMSO-d6 (Fig. 4a). The peaks due to the P5OH units at 7.58–6.60 ppm were complicatedly split owing to the presence of several conformers, which indicated that threading of the PEO resulted in frozen rotation of the P5OH π-panels. The molar ratio of EO units to P5OH was determined to be 9 from the integral values. In gel permeation chromatography (GPC) analysis, broad P5OH-derived UV absorption peaks were detected in a higher molecular weight region than the peak of uncomplexed P5OH (Fig. 4b). CO2 gas adsorption behavior derived from free P5OH was also disappeared in P5OH•PEO20k-CONHAd (Supplementary Fig. 19). These results demonstrated the existence of complexed P5OH and absence of cavity-free P5OH, thereby confirming the formation of a polyrotaxane.
Polyrotaxanes were also synthesized using THF, DMF, and acetonitrile as solvents for the capping reaction, which resulted in decreased yields and decreased numbers of threaded P5OH units (Supplementary Table 4). Effects of EO-to-P5OH ratios on the thermal properties of polyrotaxanes were evaluated by TGA (Supplementary Fig. 24). Additionally, synthesis of a polyrotaxane with P5OEt rings was attempted using the corresponding pseudopolyrotaxane prepared by the immersion method,22 which was unsuccessful because of facile de-threading (Supplementary Fig. 15). These results exemplified the critical importance of hydrogen bonds for stable complexation in organic solvents and indicated that cleavage was mediated by polar solvents, which promoted de-threading and solubilizing processes.
Thermal and mechanical properties of a high-molecular-weight polyrotaxane. A polyrotaxane with a high-molecular-weight PEO axle, P5OH•PEO500k-CONHAd (EO-to-P5OH ratio: 28), was prepared using acetone as the solvent, as well as the α-CD counterpart α-CD•PEO500k-CONHAd. From the TGA results, the 10% weight loss temperatures were determined to be 275 and 355 °C for the α-CD- and P5OH-based polyrotaxanes, respectively (Supplementary Fig. 25). The improved thermal stability of P5OH•PEO500k-CONHAd was attributed to the rigid aromatic structure of P5OH as compared with cyclodextrins, which are composed of sugar units linked via fragile acetal bonds.
DSC measurements were also performed for these polyrotaxanes and the parent PEO500k-COOH (Fig. 5a and Supplementary Fig. 22). While α-CD•PEO500k-CONHAd did not show any peaks in the region of 0–160 °C, P5OH•PEO500k-CONHAd showed two sets of endothermic/exothermic peaks at 50.7/27.2 and 60.3/42.0 °C. The peaks at higher temperatures were attributed to the melting behavior of naked parts of PEO chains and the remaining broad peaks were ascribed to EO units near P5OH rings. Importantly, P5OH•PEO500k-CONHAd did not melt upon heating over 60.3 °C, but exhibited thermoplastic behavior while retaining a solid state, which could be rationalized by the presence of multiple hydrogen bonds. Because of this thermo-responsive mechanical property, heating to 125 °C allowed the shape of P5OH•PEO500k-CONHAd to be altered, which was retained at room temperature and then was returned to the original form when it was heated again to 125 °C (Fig. 5b and Supplementary Movie 2). This shape memory property was also investigated in terms of variable-temperature PXRD patterns (Fig. 5c and Supplementary Fig. 17). At 30 °C, P5OH•PEO500k-CONHAd showed two peaks at 2θ = 19.2 and 23.4°, which disappeared between 60–80 °C in the heating scan and were retrieved between 60–40 °C in the following cooling scan. This behavior was almost the same as that of the parent PEO500k-COOH, which suggested that the temperature-dependent mechanical response can be attributed to the local melting process of the PEO chains.