Molecular Engineering of the Kinetic Barrier in Seeded Supramolecular Polymerization

Seeded supramolecular polymerization (SSP) is a method that enables the controlled synthesis of supramolecular structures. SSP often relies on structures that are capable of self-assembly by interconverting between intramolecular and intermolecular modes of hydrogen bonding, characterized by a given kinetic barrier that is typically low. The control of the polymerization process is thus limited by the propensity of the hydrogen bonds to interconvert between the intramolecular and intermolecular modes of binding. Here, we report on an engineering of the polymerization kinetic barriers by sophisticated molecular design of the building blocks involved in such SSP processes. Our designs include two types of intramolecular hydrogen-bonded rings: on one hand, a central triazine tricarboxamide moiety that prevents self-assembly due to its stable intramolecular hydrogen bonds and on the other hand, three peripheral amide groups that promote self-assembly due to their stable intermolecular hydrogen bonds. We report a series of molecules with increasing bulkiness of the peripheral side chains exhibiting increasing kinetic stability in the monomeric form. Owing to the relative height of the barrier, we were able to observe that the rate constant of seeding is not proportional to the concentration of the seeds used. Based on that, we proposed a new kinetic model in which the rate-determining step is the activation of the monomer, and we provide the detailed energy landscape of the supramolecular polymerization process. Finally, we investigated the hetero-seeding of the building blocks that shows either inhibition or triggering of the polymerization.


Materials and Methods
Materials: All Solvents and reagents were received from Acros, Aldrich TCI or Merck and were used without further purification, unless otherwise mentioned. The Benzene-1, 3, 5-tricarboxamides (BTA) model compound was synthesized according to a protocol found in literature. 1 Methods: 1 H NMR and 13 C NMR spectra were obtained on a Varian AMX400 ( 1 H: 400 MHz, 13 C: 101 MHz) spectrometer at room temperature (25 ℃). Temperature dependent 1 H NMR for self-assembly characterization were obtained on a Varian Unity Plus ( 1 H: 500 MHz, 13 C: 125 MHz) spectrometer. Mass spectra (MS) were obtained on a Waters Xevo G2 TOF spectrometer with ESI ionization. Flash column chromatography was performed on BUCHI Pure C-810 Flash system with commercial column Flash Pure ID Silica 40μm (12 g or 40 g). The UV-Vis absorption spectra were obtained on an Agilent Technologies Cary 8454 UV-Vis spectrometer with 1 cm path length quartz cuvette. The circular dichroism (CD) absorption spectra were obtained on a JASCO-815 CD spectrometer with 1 cm path length quartz cuvette.
The liquid-phase IR spectra were obtained at room temperature (25 ℃) on a Perkin-Elmer FT-IR Spectrometer 400 with a demountable liquid cell (KBr windows and PTFE spacer 1mm path length).
TEM images were taken with a CM120 microscope at 120keV acceleration voltage on plain carbon coated grids. The atomic force microscopy (AFM) imaging was performed by a PicoLE microscope (Molecular imaging) using ACAFM mode. Standard silicon nitride cantilevers (PointProbe, Nanosensors) with resonance frequency 320 kHz were used. Muscovite mica sheets (EMS) were used as a substrate. AFM samples were prepared by drop casting. DLS measurements were performed at room temperature on a Malvern Panalytical Zetasizer Ultra Red equipped with a 1 cm quartz cuvette.
Seeds preparation: All seeds are prepared by sonicating 2 mM assembled TTA 1 -TTA 4 solution for 20 s at r.t. Before the addition of TTA 4 seeds, extra MCH is used to dilute the seeds to 30 μM and add to the 30 μM TTA 4 monomer solution at the selected volume ratio. For the addition of TTA 1 -TTA 3 seeds to the 30 μM TTA 4 monomer solution, the sonicated 2 mM are directly added to the TTA 4 solution with the selected volume assuming the total volume of solution after addition is unchanged. The IR spectrum of the BTA control building block, which cannot form any intramolecular hydrogen bonds, reveals a N-H stretch centered at 3448 cm -1 in CHCl3 (non-hydrogen-bonded state). This wavenumber is higher than all three N-H stretches of TTA 0, such shift indicates the difference between intramolecularly hydrogen-bonded states and non-hydrogen-bonded states. 2 Like TTA-0, the IR spectra of TTA 1 -TTA 4 in CHCl3 ( Figure S2a  We attribute the N-H stretch centered at 3439 cm -1 to the peripheral amide groups of TTA 1 -TTA 4. Unlike TTA 0, there is no shoulder at 3434 cm -1 (proton 4) because in TTA 1 -TTA 4, this potential band is covered by the intense N-H stretches of the peripheral amide groups and therefore it is difficult to investigate this aspect by IR spectroscopy. It is also difficult to conclude on the formation of intramolecular hydrogen bonds by the peripheral amide groups of TTA 1 -TTA 4 in chloroform, even though the absence of intermolecular hydrogen bonds in chloroform is as obvious as in the BTA control, as discussed bellow.
The IR spectra of all compounds in MCH is shown in Figure S2b. The IR spectrum of the BTA control in MCH reveals an N-H stretch centered at 3241 cm -1 , this large shift (3448 cm -1 in CHCl3) towards lower wavenumbers, indicates the typical intermolecular hydrogen bonding of BTA fibers in MCH.
The infrared spectrum of TTA 0 in MCH ( Figure S2b) is similar to its spectrum in CHCl3. It reveals an N-H stretch centered at 3391 cm -1 (corresponding to the N-H stretch centered at 3386 cm -1 in CHCl3) with a shoulder at 3437 cm -1 (corresponding to the shoulder at 3434 cm -1 in CHCl3). It also reveals an N-H stretch centered at 3270 cm -1 (corresponding to the N-H stretch centered at 3256 cm -1 in CHCl3). And similarly to its 1 H NMR spectrum in CDCl3, the 1 H NMR spectrum of TTA 0 in MCH-d14 also reaveals a dedoubling of the amide peaks ( Figure S2c). So these results suggest that TTA 0 is intramolecularly hydrogen bonded in chloroform as well as in MCH (TTA 0 monomers).
On the contrary, the IR spectra of TTA 1 -TTA 4 in MCH reveals a large shifts of the N-H stretches towards lower wavenumber in comparison to their N-H stretches in chloroform ( Figure S2b). The N-H stretches attributed to the peripheral amide groups decrease from 3439 cm -1 to 3286 cm -1 . And the N-H stretches attributed to the central amide groups decrease to wavenumber close to those of the BTA fibers. For TTA 1 -TTA 3, they decrease from 3380 cm -1 to 3235 cm -1 (into a shoulder), while for TTA 4 it decreases from 3380 cm -1 to 3225 cm -1 (into a relatively sharper peak). The large shifts towards lower wavenumbers of both central amide groups and peripheral amide groups indicate that these two types of amide groups are intermolecularly hydrogen bonded in MCH (TTA 1 -TTA 4 fibers). 4. Self-assembly mechanism of TTA 1 -TTA 2.  Reduced-scale images allow visualizing thinner fibers, such as for TTA 4 in Figure S5c. The inhomogeneity of the contrast along a fiber can be explained by the presence of long flexible alkyl chains.
We assume that alkyl chains prevent determining the structure of the packing of the aromatic core of TTA 1 -TTA 4 due to the insufficient stiffness and density of the packing to reproduce the internal structure of the fiber. Here we investigate the kinetic barrier again by thermal hysteresis at a lower concentration to compare the kinetic barrier between TTA 4 and TTA 3. As Figure S7 shows, the ∆ e of TTA 4 is 21 K. This is much higher than the ∆ e of TTA 3 which is only 9 K. This is because at the lower concentration influence of fiber fragmentation on nucleation time is reduced and the main contribution comes from the primary nucleation which reflects the height of the real kinetic barrier. As the hysteresis comes from the nucleation time, TTA 4 which has a higher kinetic barrier shows a bigger ∆ e . For more details, the reduction of influence from fiber fragmentation on nucleation time at lower concentrations can be explained below.

Heating and cooling of TTA 1 -TTA 4 solutions followed by UV-Vis spectroscopy
During the nucleation phase, the process is proposed as follows: Where M * is the activated monomer, S is the fiber with length , c is the size of the nucleus, n * is the nucleation rate constant, the nucleation, ma * is the elongation rate constant, and f is the fragmentation rate constant. Here we ignore the reversibility. Because during the nucleation phase, the rate of associating M * to S is very low due to the low concentration of S , we can also assume [M * ] = is a constant which is concentration-independent. Therefore we can replace M * with M in the eq S1 and eq S2.
Then the reduction of the influence of fragmentation at lower concentrations can be explained by the model proposed by Kym Eden et al. 4 They assume primary nucleation (eq S4), fibril elongation (eq S5), and fragmentation (eq S6) happens during the self-assembly process. Because replacing M * with M doesn't change the equation deduced by fragmentation, eq S6 is unchanged compared with eq S3.
Where M is the trapped monomer, S is the fiber with length , c is the size of the nucleus, n = c n * is the nucleation rate constant, the nucleation, ma = ma * is the elongation rate constant and f is the fragmentation rate constant which is the same as the one in eq S3.
Given the condition ∝ [M] tot c −1 , we found the ratio of the first term of eq S7 to its second term is Where y is CD value, t is time, a and b are parameters. By fitting this integrated equation to the data above, we can get + . For the control group, the TTA 1 seeding group and the TTA 2 seeding group because of the poor correlation coefficients at first 16 min ( Figure S12a, c, e), no reliable size can be measured by DLS.

Seeded and hetero-seeded supramolecular polymerization of TTA 4 followed DLS
However, these poor correlation coefficients can indirectly support that TTA 1 and TTA 2 cannot trigger the self-assembly of TTA 4 monomer. After 32 min correlation coefficients of these three groups become good enough to get reliable DLS results ( Figure S12g). At this time control group shows a peak around 19 seeding group, due to the hetero-seeding experiment followed by CD (Figure 7), there is no assembled state formed after 32 min. Therefore the peaks of these two groups are probably due to the partial clustering of TTA 1 and TTA 2 seeds.
For the TTA 3 seeding group and the TTA 4 seeding group the correlation coefficient is always good enough to get reliable DLS results ( Figure S12a, c, e, g). The DLS results show that after the seeding of TTA 3 and TTA 4, the size increases gradually from 1 min to 16 min ( Figure S12b, d, f), and stays at that size after 32 min ( Figure S12g). These DLS results are consistent with the CD results that both TTA 4 and TTA 3 seeds can trigger the self-assembly of TTA 4 monomer. The extra peak that occurred at 32 min for TTA 3 seeding group can be probably attributed to the partial clustering of the remaining TTA 3 seeds just like the case in TTA 1 and TTA 2 seeding groups.
14. Heating curve of TTA 1 -TTA 4 fitted to a cooperative model. The significantly lower ∆ e of TTA 4 compared to TTA 1 -TTA 3 is probably due to the larger R group.
It forces the central amide groups to rotate at a larger angle to form intermolecular hydrogen bonds and this bond is stronger than the one in TTA 1 -TTA 3. This is consistent with the IR spectra of TTA 4in MCH, showing that the N-H stretch of the central amide groups is a peak centered at lower wavenumbers, while the N-H stretch of the central amide groups of TTA 1 -TTA 3 is a shoulder at relatively higher wavenumbers ( Figure S2b). Although the ∆ e of TTA 4 is significantly lower, its large R group can create a large entropy penalty after self-assembly. Thus TTA 4 fibers are easier to fragment than TTA 1 -TTA 3 fibers.  N 2 ,N 4 ,N 6 -tri((S)-nonan-2-yl)-1,3,5-triazine-2,4,6-