Preprogrammed assembly of supramolecular polymer networks via the controlled disassembly of a metastable rotaxane

In our daily life, some of the most valuable commodities are preprogrammed or preassembled by a manufacturer; the end-user puts together the final product and gathers properties or function as desired. Here, we present a chemical approach to preassembled materials, namely supramolecular polymer networks (SPNs), which wait for an operator’s command to organize autonomously. In this prototypical system, the controlled disassembly of a metastable interlocked molecule (rotaxane) liberates an active species to the medium. This species crosslinks a ring-containing polymer and assembles with a reporting macrocycle to produce colorful SPNs. We demonstrate that by using identical preprogrammed systems, one can access multiple supramolecular polymer networks with different degrees of fluidity (μ* = 2.5 to 624 Pa s-1) and color, all as desired by the end-user.


Supplementary methods
General All commercially available chemicals were purchased from Sigma-Aldrich, Tokyo Chemical Industry (TCI), and Oakwood Chemical, and used as received. Dry dichloromethane (CH2Cl2) was collected from an Inert PureSolv MD5 purification system, whereas acetonitrile (CH3CN) and chloroform (CHCl3) were freshly dried with activated 3 Å molecular sieves. Deuterated solvents (CD3CN and CDCl3) were purchased from Cambridge Isotope Laboratories and freshly dried using activated 3 Å molecular sieves.
Flash column chromatography was carried out using SiliCycle (230-400 mesh) silica gel as the stationary phase. Nuclear magnetic resonance (NMR) experiments were recorded on Bruker AVIII HD 400 MHz and Bruker Avance 400 MHz spectrometers; 1 H and 13 C NMR chemical shifts (δ) are given in parts per million (ppm) relative to tetramethylsilane as referenced with the residual solvent signal.
J values are reported in Hz, and signal multiplicity is denoted as s (singlet), d (doublet), t (triplet), dd (doublet of doublet), m (multiplet), and br (broad signal). UV-vis spectra were recorded on a Cary 5000 UV-vis-NIR spectrometer, employing 1 mm pathlength quartz cuvettes. Electrospray ionization -high-resolution mass spectra (ESI-HRMS) were recorded on an ESI-TOF Waters Micromass LCT spectrometer. MALDI-TOF mass spectrometry experiments were performed using a Bruker Autoflex.
GPC data were collected on a Malvern Omnisec chromatographer equipped with two T6000M columns arranged in series; the experiment was performed at a flow rate of 1 mL min -1 at 35 °C in THF, and using a refractive index detector. Rheology tests were carried out on Anton Paar Modular Compact Rheometer MCR 502 with a cone plate geometry (diameter 25 mm, cone angle 1º).
The resulting isotherms were fitted by a nonlinear least-squares method using a 1:1 global fitting model (Nelder-Mead method) in the BindFit platform. [  with H1c, H1d, H1e, H1g, and H1h, respectively, suggesting that DB24C8 sit on the DBA moieties of 1 4+ , as well as [3]PR. The assembly/disassembly motion of DN38C10 and BIPY core of 1 4+ became slower in [4]PR than that in [2]PR, probably because the speed was limited by the slow assembly/disassembly of DB24C8 and DBA moieties of both ends of 1 4+ . Finally, ESI-HRMS showed the molecular ion [1    appeared at an upfield position (6.81 ppm) after 1 d of heating, followed by another peak appearing at a further upfield position (6.77 ppm) on the 4 th heating day. The appearance of new peaks at the upfield positions indicates that DBA lost electron-rich 22C6 rings [6] . Focusing on their intensity changes over 30 d heating, while the original Hb [3] peak monotonically decreased to almost zero, the first appeared peak increased to reach a maximum on the 10 th heating day and decreased after that; on the other hand, the second new signal grew to be dominant. These characteristic intensity changes mean that the original Hb [3] peak was converted into the first appeared peak followed by another conversion into the second peak. Considering the order of appearance and positions of the peaks, and overall intensity changes, it is reasonable to explain the observed peak changes as the following process: [ Figure 36).

Supplementary note 1
For a better understanding, we carried out two quantitative analyses of the dissociation process of [3]MSR. We first estimated the dissociation ratio of 22C6 rings depending on the heating time. The dissociation process of 22C6 was tracked by the peak shift from the complex form (c) to the uncomplexed (uc) form in 1 H NMR spectra (Supplementary Figure 32b). We calculated the dissociation ratio using the integration of proton Hc from Equation (11), which was derived as follows:

S42
The dissociation ratio of 22C6 after n days of heating was defined as: Where,  To calculate the dissociation ratio using the integrations from each zone, we rearranged Equation (3) to Equation (11) by applying Equation (4) Equation (11) We calculated the dissociation ratio from Equation (11) (17) The integration of [3]MSR can be described as Equation (18) by combining Equation (16) and (17). To estimate the reliability of these two analyses, we calculated the dissociation ratio of 22C6 differently from Equitation (11) by using the obtained existence ratio of [3], [2]MSR, and free 1 4+ .
The [3]MSR equips with two 22C6 rings and releases one of 22C6 converting into [2]MSR, which releases the other 22C6 providing free 1 4+ . This process means that the equivalence of dissociated 22C6 is one and two to [2]MSR and free 1 4+ , respectively. Based on this relationship, the dissociation ratio of 22C6 is estimated by Equation (20):

…Equation (20)
The plots of the dissociation ratio calculated from Equation (20) agreed with those from Equation (11) indicating that these quantitative analyses are highly reliable (Supplementary Figure 37).

Stability test of dissociation process of [3]MSR
To confirm that the dissociation process of [3]MSR is irreversible, a solution of [3]MSR in dry CD3CN/CDCl3 (1:1, v/v, 3.5 mM) was heated at 70 ºC for 12 days and then cooled down and left at r.t for 20 days. Subsequently, the sample was heated at 70 ºC for another 22 days (30 days of heating in total) and then stored at r.t for 1 year. After each treatment, the sample was analyzed by 1 H NMR spectroscopy (Supplementary Figure 39). In the collected 1 H NMR spectra, we observed dissociation progressed in the heating process but did not observe any significant changes during the storage period at room temperature. This means the thermal-triggered dissociation process is irreversible and can stop and restart by turning off/on the heating.
On the other hand, a sample containing dibenzylammonium (DBA) hexafluorophosphate and cis/trans-22C6 in equimolar amounts (5 × 10 -3 M), in C2D2Cl4, was analized by 1 H NMR spectroscopy ( Figure S38e); the experiment showed no self-assembly, suggesting that the 22C6 ring cannot slip over the phenyl rings contained on DBA at ambient conditions. Even after heating at 130 °C for 7 d, we did not detect diagnostic signals for rotaxane assembly ( Figure S38f). Likewise the dissociation studies of [3]MSR, The samples were heated at 70 ºC up to 30 d and periodically monitored by 1 H NMR spectroscopy.
The mixtures of [3]MSR and free rings, DN38C10 and/or DB24C8, did not produce any pseudorotaxanes. When heating the systems, however, all ring-exchange systems toward [2], [3], or [4]pseudorotaxanes were successfully performed, involving the disassembly of [3]MSR and the following self-assembly with prepared rings (See "Ring-exchange from [3]metastable rotaxane to [4]pseudorotaxane" section for detailed characterization).

Ring-exchange from [3]MSR to [4]pseudorotaxane
The stopper function of [3]MSR preventing 1 4+ from forming PRs with free rings was confirmed by 1 H NMR studies. A solution containing [3]MSR (3.5 mM), DN38C10, and DB24C8 (1:1:2 mol ratio) in CD3CN/CDCl3 (1:1, v/v) was prepared. In the collected 1 H NMR spectrum, we observed all peaks belonging to 3 components ([3]MSR, DN38C10, and DB24C8) overlapped, which indicates all components co-exist without forming any pseudorotaxanes (Supplementary Figure 47). However, protons H1a and H1b shifted upfield ( = 0.03 and 0.08 ppm, respectively), which suggests that the naphthalene moieties of DN38C10 externally interact with the bipyridinium core of [3]MSR. [10], [11] In fact, the prepared NMR solution showed pale pinkish color due to this external interaction of DN38C10 (Supplementary Figure 46). This NMR sample was stored at room temperature for 30 d and we observed no significant changes in the 1 H NMR spectrum over this period (Supplementary Figure   48). This result indicates the sample is very stable, with no reaction occurring at room temperature. Next, the system was heated to perform the conversion from [3]MSR to [4]PR via ring-exchange on 1 4+ . We prepared a solution of Analysis of color change over transformation from [3]MSR to [4]pseudorotaxane The heating-time-dependent color change of the preprogrammed system was investigated by UV-vis spectroscopy. Before preprogrammed system prepared the previous section, we started the analysis of a simple system containing DB24C8 instead of poly(DB24C8) for a fundamental understanding of color change caused by ring-exchange from metastable-to-pseudorotaxane.
As a simple programmed system, we prepared seven independent solutions containing [3] Figure 52a). We did not observe any viscosity increase in the heated samples because this simple system did not contain poly(DB24C8) S63 and could not form a supramolecular network. All seven solutions were diluted to 3 mM based on [3]MSR concentration and analyzed by UV-vis spectroscopy.

Supplementary note 2
The absorbance of non-heated control at 486 nm was higher than that of pure [3]MSR because of the external complex formation. Depending on the heating time, the absorbance at 486 nm representing the charge-transfer interaction of BIPY core ⸦ naphthalene moieties in pseudorotaxane structure gradually increased to give a summit, although the summit at 486 nm was not observed before heating (Supplementary Figure 52b). To confirm the observed absorbance increase is derived from [4]PR formation, the sample heated for 30 d was evaporated and then prepared as a solution in CD3CN/CDCl3 (1:1, v/v) for 1 H NMR analysis. The collected 1 H NMR data (Supplementary Figure 53) showed both dissociation of [3]MSR and formation of [4]PR, indicating that the thermal-triggered color change was caused by ring-exchange from [3]MSR to [4]PR.
To confirm the observed heating-time-dependent solution color change was performed in a programmed manner, we conducted a quantitative investigation into the relationship between collected absorbance data and the existence ratio of [3]MSR and its related dissociated species shown in   Based on the results of the basic programmed system using non-polymeric DB24C8, the preprogrammed supramolecular network systems containing poly(DB24C8) prepared in the previous section were analyzed identically by UV-vis spectroscopy. We heated 8 samples of system 1 at 70 ºC for 0 (control), 4, 6, 10, 14, 18, 22, and 30 days, separately, and also heated 6 samples of system 2 for 0 (control), 5, 10, 16, 26, and 30 days, separately. In both systems, the initial samples without heating were free-flowing solutions with yellowish-orange color, but they became tough red gel after 30-days of heating. Besides, each system showed gradual changes in terms of both color changes and viscosity increases depending on heating time. For the quantitative analysis of color change, we diluted the S65 system to solutions (1.7 mM for system 1 and 3.0 mM for system 2), measured their absorbance spectra, and then plotted the collected absorbance increase ratio calculated from Equation (24)  DB24C8 = 1: 1: 2 in molar ratio) (red), system 1 (green), and system 2 (blue).

Synthesis and characterization of poly(DB24C8)
Polymer poly(DB24C8) was prepared based on a reported procedure [6] and characterized by 1 H NMR spectroscopy, GPC and MALDI-TOF mass spectrometry (see Supplementary Figure
Eight and six independent samples were prepared for system 1 and system 2, respectively. The prepared samples in both systems 1 and 2 were free-flowing solutions with light orange color, which is attributed to weak CT from the electron-rich DN38C10 ring to the BIPY unit of 1 4+ through an unthreaded geometry.

Analysis of viscosity change over transformation from [3]MSR to [4]pseudorotaxane
To probe the thermal-triggered viscosity changes of the preprogrammed systems, systems 1 and 2 prepared in the previous section were analyzed by a rheometer. We heated eight samples of system 1 for 0 (control), 4, 6, 10, 14, 18, 22, and 30 days and six samples of system 2 for 0 (control), 10, 16, 20, 26, and 30 days, separately. In both systems, the starting solutions were low viscosity, free-flowing liquids before heating. As the samples heated, we observed an increase in viscosity and transformation to a gel on the 14th day of heating in system 1 and the 10th day of heating in system 2. The rheological analysis of the preprogrammed systems was conducted by the dynamic oscillatory shear measurements To get quantitative insights on viscosity increase depending on the heating time, the obtained complex viscosity at 8.11 rad s -1 is plotted against the expected existence ratio of free 1 4+ calculated from Equation (19) in Supplementary Figure 63. The obtained plots show a strong correlation between complex viscosity and the existence ratio of free 1 4+ in both system 1 and 2, suggesting that the thermal-triggered viscosity increases are caused by the released 1 4+ . This is a reasonable result because 1 4+ is the only species that can work as a cross-linker for poly(DB24C8) among [3]MSR and its dissociated species.
To get a deeper understanding of the preprogrammed viscosity increase, we prepared control samples that represent the ideal terminal states of the programmed system where all [3] 2. This means the thermally-triggered viscosity increases of the preprogrammed systems proceeds toward their fully cross-linked states. In summary, the programmed system can increase its viscosity toward that of a fully cross-linked state with longer heating time.