The Role of Sodium Cations and Covalent Decoration in the Dissociation of Supramolecular Polymers and a Rational Strategy to Improve Their Stability

ABSTRACT

Heterotypic supramolecular hydrogels are soft materials obtained by mixing at least two monomers in aqueous solution that self‐assemble via hydrogen bonds (HBs) into polymeric fibers. Despite attempts to obtain stable hydrogels, some systems precipitate over time. In this work, we uncover the role of sodium cations and covalent modifications on the stability of three supramolecular polymers based on cyanuric acid (Cy) and triaminopyrimidine (TAP). Our computational experiments show that sodium cations penetrate the polymer and break the inner HBs, thus perturbing the stability of the structure. The longer and more flexible hexanoic acid (Co6) side group–compared to the butanoic acid (Co4) group on the Cy moiety– promotes more favorable hydrophobic close contacts with each other. Based on these results, we designed in silico a system based on CyCo6 and a triaminotriazine (TAT) moiety modified with a succinate group (sa). In this new polymer, the TATsa molecules effectively coordinate the cations on the outer surface of the polymer, avoiding their entry into the channel and thereby improving the HB strength of the rosettes and the stability of the polymer. Overall, our findings provide an understanding of the physicochemical factors that cause the precipitation of supramolecular hydrogels.

Molecular dynamics (MD) experiments reveal two critical factors promoting the instability and precipitation of supramolecular hydrogels based on hydrogen‐bonded rosettes: cation penetration into the polymer channel and the absence of hydrophobic side chains that impose a helical twist on the polymer. These insights enable the design of a supramolecular system that bypasses these destabilizing factors, ensuring structural integrity in aqueous solution.

1. Introduction

Hydrogels are soft materials, neither solid nor liquid, that are made of cross‐linked polymeric structures with the capacity of retaining large quantities of water. Heterotypic supramolecular hydrogels are obtained when at least two different monomers are mixed in aqueous solution and the resulting polymeric network is supported by noncovalent forces [1]. Melamine [2] (M) and its derivatives [3, 4]. have been widely used to obtain this type of hydrogel. The increasing interest in these materials stems from their potential applications across several disciplines. For example, Wang et al. [5] have used an M‐based hydrogel to remove uranium(VI) from water, while Rao et al. have employed self‐assembled systems of barbituric acid derivatives to detect Hg²⁺ ions in aqueous solution [6]. Nandi et al. [7] have also used M as a cogelator to finely tune the morphological structure of the hydrogel along with their photoluminescence properties. Experiments with rosette nanotubes [8, 9] suggested they could be used as vehicles to encapsulate and protect therapeutic drugs, enabling their controlled release. In the field of medicine, a hydrogel based on rosette nanotubes covalently modified with a cytocompatible peptide was used as biomimetic scaffolds for tissue engineering [10]. The same rosette‐based supramolecular structure was also employed to improve the synthesis of monodispersive gold nanoparticles [11].

Hud and coworkers [12] have synthesized a series of heterotypic hydrogels by mixing triaminopyrimidine (TAP) with cyanuric acid (Cy) substituted with either a hexanoic acid (Co6) or butanoic acid (Co4) side chains, CyCo6 and CyCo4, respectively. It has been shown that the stability of the gels over time depends strongly on the covalent modification of one of the moieties. While an equimolar mixture of TAP:CyCo6 forms a stable hydrogel for over six months, the TAP:CyCo4 mixture is unable to gel. Furthermore, the authors observed gelation and greater stability as more CyCo6 is added to the TAP:CyCo4 binary mixture. Although supramolecular polymerization of a system can be induced by increasing the ionic strength of the medium [13], as was also demonstrated with the TAP:CyCo4 system [12], the intriguing physicochemical factors that cause the precipitation of the hydrogels has not yet been disclosed. We have demonstrated recently that metal ion coordination interactions can compete with and hinder hydrogen bond (HB) formation in supramolecular polymers [14]. For instance, depolymerization effects triggered by metal ion coordination were reported by Jang et al. [15] upon the addition of Cu²⁺ or Ag⁺ to a host‐guest fibrous system. Noncovalent interactions imposed by the side chains are also fundamental for the stability of the structure [14, 16].

In this work, we analyze the stability of three hydrogen‐bonded polymers: TAP:CyCo4, TAP:CyCo6, and a third variant of Cy substituted with 4‐methylhexanoic acid [17, 18] (4MeCo6), TAP:Cy4MeCo6 (see Scheme 1), using molecular dynamics (MD) to understand the physicochemical factors that cause the dissociation of the polymers and further precipitation of the hydrogels. We also investigate why the addition of CyCo6 to TAP:CyCo4 increases the lifetime of the hydrogel. Based on our found understanding of the phenomenon, we designed a supramolecular polymer that avoids the entry of sodium counterions into the channel, thus improving the stability of the HB network.

SCHEME 1.

Molecular structures of 2,4,6‐TAP and cyanuric acid (Cy‐R1, R1 = H) substituted with butanoic acid (Co4), hexanoic acid (Co6), and 4‐methylhexanoic acid (4MeCo6). TAP and Cy‐R1 form the hydrogen‐bonded hexad (rosette) shown below.

2. Computational Details

2.1. Molecular Dynamics (MD)

All MD simulations were conducted using GPU‐accelerated Amber22 software package [19], adopting periodic boundary conditions. The GAFF2 force field was employed to model the monomers, and their atomic charges were calculated with the RESP method [20] using the Gaussian 16 software [21]. This force field was tested on a similar polymer in solution, and the results have shown great consistency with the experimental observation [14].

The initial structures for the MD simulations were constructed following a previous protocol [14]. The monomers were then accordingly modified by adding the R groups on the Cy moiety, as shown in Scheme 1. The side chains (Co4, Co6 and 4MeCo6) were further optimized at the PM7 semiempirical level of theory while keeping the TAP₃Cy₃ hexad core frozen. Then, the modified hexads were used as input to extend the polymer up to a total of 20 rosettes using a Python script (see full script and usage examples in the GitHub repository at https://github.com/LeonardoGomezChavez/Nanocables.git). The distance and the helical twist between every two adjacent rosettes were set at 3.4 Å and 30°, respectively, according to reported experiments and computations [14]. All polymers were then solvated in truncated octahedral simulation cells with around 13000 OPC3 water molecules. To neutralize the carboxylate groups, sodium ions were used as counterions (see octahedral boxes in Figure S1).

Equilibration of the polymers began with a short minimization step, followed by a 500 ps heating phase and a 1500 ps equilibration period under constant pressure at 300 K. A short 5 ns simulation was then performed in the NPT ensemble to avoid further distortions, with temperature control assisted by a Langevin thermostat. The pressure was stabilized at 1 atm by the Langevin barostat using a relaxation time of 2 ps. Following this step, production simulations were performed in the NVT ensemble for 100 ns, with temperature control provided by a Berendsen thermostat. The SHAKE algorithm was used to constrain bonds involving hydrogen atoms, allowing a time step of 2 fs. Hydrogen bonds (HBs), Radial Pair Distribution Functions (RDF) and Root Mean Square Fluctuation (RMSF) analyses were performed with the cpptraj module implemented in Amber22.

All figures were created with VMD [22], Marvin [23], and SketchUp.

2.2. Density Functional Theory

All isolated monomers and complexes were optimized at the M06‐2X/6–311++G(d,p) level of theory using the Gaussian 16 suite of programs [21]. This functional has recently exhibited great performance in hydrogen‐bonded systems [24], and it has demonstrated good agreement with experimental results [25]. The minimum energy nature of the optimized structures was verified using the vibrational frequency analysis. Geometry optimizations were also carried out in water at the same level of theory by employing the polarizable continuum model [26] (PCM). All interaction energies (hydrogen bond energies ∆E _HB for dimers, and coordination energies ∆E _coor for coordination complexes with Na⁺) were computed as the difference between the energy of the considered complex and the energies of the corresponding monomers with the structure they have in the complex.

Molecular interactions were analyzed within the framework of the quantum theory of atoms in molecules (QTAIM). Local properties at the bond critical points (BCPs) were computed using the AIMAll program [27] and molecular graphs were generated with Multiwfn [28] and visualized with VMD [22]. Regions of noncovalent interactions were analyzed with the reduced density gradient [29] (RDG) approach. All wave functions were obtained at the same level of theory.

3. Results and Discussion

3.1. Binary Mixtures

3.1.1. Structure and Mobility

Figure 1 shows snapshots of three supramolecular systems simulated for 100 ns that were studied in previous experiments [18, 30]. Although our production simulations might suggest the structures are stable in solution, we subsequently demonstrate why gelation is not observed for the TAP:CyCo4 mixture. The structural mobility of the polymers was evaluated by calculating two trajectory analyses: the average RMSF for each molecule (see Figure S2) and the dihedral angle distributions of CyCo4, CyCo6, and Cy4MeCo6 side chains (see Figure S3). The TAP:Cy4MeCo6 polymer exhibits the lowest fluctuations, which aligns with its highly ordered structure. Conversely, TAP:CyCo4 and TAP:CyCo6 display a wavy structure and their monomers show greater fluctuations (see Figures 1 and S2).

FIGURE 1.

Snapshots of TAP:Cy‐R1 supramolecular polymers after 100 ns. Not all sodium cations are shown for clarity.

The mobility of the ‐Co4, ‐Co6, and ‐4MeCo6 side chains was then evaluated by looking at their dihedral angle distributions. Given that the dihedral angles of the ‐Co4 side chains are 180° (as shown in the histograms of Figure S3), the side chain is predominantly extended. In contrast, the ‐Co6, and ‐4MeCo6 side chains, two of their dihedral angles show distinct peaks at 60°, which indicates that these side chains are more flexible and capable of folding. Furthermore, four helical parameters that define the polymer structure [18] are also observed: the distance between rosettes (rise), the angle of rotation between rosettes (twist), the displacement between rosettes (slide) and the angle formed by the planes of two adjacent rosettes (roll), as outlined in Figure S4. These helical movements are more pronounced in the TAP:CyCo4 and TAP:CyCo6 polymers. Other types of movements between monomers such as propeller twists, buckles and tilts, like those observed in DNA base pairs [31], are also observed.

3.1.2. Coordination and Hydrogen Bonding Interactions

The sodium cations and water molecules (purple spheres and light blue molecules in Figure 1, respectively) that enter the polymer channel inevitably perturb the inner HB system. Accordingly, the TAP:CyCo4 system is more prone to disruption of its HB network due to the larger number of Na⁺ cations and H₂O molecules within the channel. Therefore, this could explain its instability in forming a hydrogel.

By looking at the radial pair distribution functions (RDF) between Na⁺ and the O and N atoms, the probability of finding cations near the O and N atoms within the channel is even higher than that of the carboxylate groups (see Figure 2). The degree of solvation of the atoms that are exposed to the solvent were also analyzed between water molecules and the O and N atoms (see Figure S5). This analysis shows that the carboxylate groups are more solvated than the carbonyl and amine groups exposed to the solvent.

FIGURE 2.

Radial pair distribution function (RDF) between sodium cations and highlighted atoms of TAP and Cy‐R1 molecules. The data were collected at the 100th nanosecond.

The D − H∙∙∙A HBs that stabilize the rosette structures were analyzed using a geometric criterion based on the distance between the donor (D) and acceptor (A) atoms and the ∠DHA angle. The cutoff distance between the D and A atoms was set at 3.5 Å and the cutoff for the angle was set at 135°. These analyses were performed over the last nanosecond of the production simulation, and the results were plotted in Figure 3. The rosettes are assembled via three distinct HBs, as illustrated in Scheme 1. As shown in Figure 3, green values correspond to the outer HBs, red values to the middle HBs, and black values to the inner HBs. Considering the polymer consists of 20 rosettes, with each rosette comprising 18 HBs (six of each type), the maximum possible number of HBs of each type within the polymer is 120. The middle HBs reach almost the maximum observed count in all three systems, and they also show very low fluctuations. Conversely, the inner and outer HBs show large fluctuations. Table 1 summarizes the weighted average of occupancy percentages (Equation 1) and the average number of HBs per picosecond. The occupancy percentage (HB%) is defined as the fraction over the total simulation time in which the geometric criteria are satisfied, and it can be used as a measure of the HBs strength [14, 32].

$$\overline{X}=\frac{\sum x_i\times N_i}{N_T}$$ (1)

FIGURE 3.

Number of HBs versus time for TAP:Cy‐R1 (R1 = Co4, Co6, 4MeCo6).

TABLE 1.

Weighted average of occupancy percentages (HB%) and average number of HBs per picosecond (N°) after 100 ns for TAP:Cy‐R1 polymers. See full data set in Table S1.

R1	In a		Mid b		Out c
	HB%	N°	HB%	N°	HB%	N°
Co4	67	91.2	100	119.5	82	107.5
Co6	68	93.4	99	111.0	87	104.4
4MeCo6	67	100.8	99	117.7	92	108.4

^a Inner interactions: N_TAP − H∙∙∙O_2,Cy‐R1.

^b Middle interactions: N_1/3,Cy‐R1 − H∙∙∙N_TAP.

^c Outer interactions: N_TAP − H∙∙∙O_4/6,Cy‐R1.

In this study, x _i represents the maximum occupancy percentage for each frequency interval, N _i is the number of data points within the intervals, and N _T is the total number of data points. Values of Table 1 indicates that the strongest interaction is the middle one (with largest HB%), followed by the outer and the inner ones. In contrast to a previous assumption [18] that the capture of cations and water molecules within the stacks contributes to the stability of the assemblies, our results demonstrate there are fewer inner HBs (see Figure 3) because they are directly affected by the presence of these species in the channel.

To further prove the stability of the inner HBs is weakened by the presence of cations, we extracted a reduced system from the MD simulation as shown in panel B of Figure 4a. The (TAP···Cy)@Na⁺ coordination complex was then optimized at the M06‐2X/6–311++G(d,p) level of theory in aqueous solution (Figure 4b). For comparison, we also optimized a complex without coordination: TAP···Cy (see also Figure 4b). As can be seen in Figure 4b, the Na⁺ cation displays two coordination interactions, one with the carbonyl group of Cy and another one with the amine group of TAP. These interactions distort the hydrogen‐bonded structure and thus weaken the inner HB. The electron density at the bond critical point (BCP) of the inner HB decreases from 0.023 a.u. in TAP···Cy to 0.007 in the (TAP···Cy)@Na⁺ coordination complex. The ΔE _HB interaction energy is also less stabilizing, going from –13.3 to –10.7 kcal mol−¹, respectively.

FIGURE 4.

(a) Molecular structure of a reduced system (B) extracted from the MD runs according to panel A. (b) Optimized structures in aqueous solution (M06‐2X/6–311++G(d,p) level) of structure B ((TAP···Cy)@Na⁺), and TAP···Cy dimer without sodium. Hydrogen bond (ΔE _HB) and coordination (ΔE _coor) energies are in kcal mol−¹. RDG surfaces superimposed on the molecular graphs are also shown. The density at the BCPs (in a.u.) is indicated in parenthesis.

Finally, we identified two other interactions that are not as relevant as those that maintain the rosette structure. The first corresponds to the HBs between the carboxylate groups and the ‐NH₂ amine groups of TAP exposed to the solvent (N_TAP–H···O_Cy‐R1), which are depicted in Figure 3 with blue lines. The second interaction is an HB system between the ‐NH₂ amine groups of the channel (N_TAP–H···N_TAP) from one rosette to another one. These HBs are also weakened when the Na⁺ cation enters the cavity (see Supporting Discussion 1 and Table S2).

3.1.3. Mechanisms of Ion Migration

We observed that sodium cations can access the channel via two pathways: through the terminal cavities that are exposed to the solvent (see Figures S6 and S7 in Supporting Discussion 1); and through gaps created by the lateral sliding of adjacent rosettes, which also exposes the inner channel to the solvent (see also Figure S8). Given that atomic force microscopy (AFM) images [18] of TAP:CyCo6 and TAP:Cy4MeCo6 polymers have shown extensively long threads, sodium ion migration through the ends is expected to have a negligible impact. To avoid the dangling ends of the polymers within our simulation boxes and identify the dominant mechanism, we simulated an infinite TAP:CyCo6 polymer using periodic boundary conditions, as shown in Figure 5 (see a detailed comparison between the truncated and infinite polymer in Figure S9). After 100 ns, four sodium cations, along with their hydration water molecules, migrated into the polymer channel. Hence, this experiment demonstrates that sodium cations can easily infiltrate into the polymer due to the sliding of the rosettes.

FIGURE 5.

Snapshots of the infinite TAP:CyCo6 polymer at 1 and 100 ns. Water molecules inside the channel were omitted for clarity.

3.1.4. Side Chain Interactions

One of the most important structural features of this type of supramolecular polymer is the helical twist between rosettes [18, 33]. The importance of this helical parameter arises from the potential contacts between the side chains. To track the helical twist, we analyzed the RDF between the N atoms to which the side chains are attached, as shown in Figure 6a,b. We then compared the position of the peaks with the RDF value of two reference structures with perfect twists of 30° and 60° (Figure 6c), represented by the green and magenta lines, respectively (see also Figure S10). The initial reference structures display sharp peaks at 4.95 Å (coincident with a perfect twist of 30°), and ∼7 Å (coincident with a perfect twist of 60°). The TAP:CyCo4 polymer, which does not form a hydrogel, shows a broad peak at around 7 Å and a small shoulder at 5 Å. In contrast, the TAP:CyCo6 polymer shows two broad peaks: one at ∼5 Å (equivalent to a helical twist of 30°) and a smaller peak at ∼7 Å (equivalent to a helical twist of 60°). The third polymer, TAP:Cy4MeCo6, shows only one narrow peak coincident with a twist of 30°. These results reveal three unique structural behaviors among the polymers. Since the helical twist of TAP:CyCo4 is more populated at 60° (indicated by a more pronounced peak at 7 Å), the side chains of the CyCo4 monomers are distant from each other and unable to interact. Conversely, the TAP:Cy4MeCo6 shows an almost perfect helical twist of 30°, which is in good agreement with the experimental data [18].

FIGURE 6.

(a) Molecular structure of CyCo4. The highlighted N atom (green) was used as a reference point to analyze the helical twist between the stacks. (b) Structure of the initial configuration of TAP‐CyCo4 with a helical twist of 30°. The interatomic distance between the highlighted N atoms is reported in Å. (c) Radial pair distribution functions between highlighted N atoms. Green and magenta lines correspond reference structures with perfect helical twist.

Regarding the TAP:CyCo6, the peaks in the RDF indicate a dual behavior. A detailed inspection of the MD trajectories reveals that the rosettes change their twist conformation from 30° to 60° and vice versa within short periods of time. For instance, Figure S11 shows two rosettes from the polymer that change their helical twist from 60° to 30° within 1 ns. This type of conformational change was verified in several parts of the polymers and throughout the trajectory, which is also consistent with the large fluctuations shown in Figure S2. To further evaluate whether the side chains interact with each other, we studied a reduced system taken from the production dynamics, as shown in Figure 7 (see panels A and B). Panel B from Figure 7 shows two CyCo6 molecules from two adjacent rosettes. We then analyzed a smaller analog by truncating the side chains, and saturating the C atoms with hydrogen, without modifying the original MD coordinates. The resulting structure was optimized in aqueous solution at the M06‐2X/6–311++G(d,p) level of theory, and the molecular interactions were analyzed using the QTAIM and RDG frameworks (see panel C in Figure 7). The ΔE _int interaction energy (PCM) of the reduced dimer is stabilizing (–1.8 kcal mol−¹). Importantly, the DFT optimized structure does not differ significantly from the structure obtained by MD. Besides, as can be seen in panel A (Figure 7) the three pairs of side chains display nearly the same interaction pattern: one chain is extended while the other one is folded. In panel C (Figure 7) two types of bond paths are recognized, three bond paths connecting H atoms and two bond paths between H and O atoms. The H···H contacts are associated with van der Waals forces (they also show green RDG surfaces), and the H···O contacts are associated with weak C─H···O HBs.

FIGURE 7.

Snapshot of TAP:CyCo6 after 100 ns. Panels A and B show the analyzed section. Panel C shows a reduced system of Co6 groups (CH₃CH₂CH₂COO–) optimized at M06‐2X/6–311++G(d,p) level in aqueous solution. RDG isosurfaces superimposed on the molecular graphs are also shown. Interaction energy ΔE _int is shown between brackets in kcal mol−¹.

Although this contact between two carboxylate groups may seem weak, the set of several contacts along the polymer is additive (like a Velcro system) and effectively helps to stabilize the polymer. Therefore, the fact that no gelation is observed when equimolar quantities of TAP and CyCo4 are mixed is based on two reasons: first, sodium cations penetrate the channel of the polymer and they weaken the inner HBs, and second, the lack of effective hydrophobic interactions between the side chains contributes to the instability of the polymer structure.

3.2. Ternary Mixtures

Hud and coworkers have demonstrated that gelation of CyCo4 and TAP is observed only when CyCo6 is added as a cogelator to the mixture [12]. In addition, the lifetime of the hydrogel increases as the ratio CyCo6/CyCo4 increases. For instance, the TAP:CyCo4:CyCo6 hydrogels with the ratios: 1:0.8:0.2, 1:0.5:0.5, and 1:0.33:0.67 precipitate after some minutes, 30 min and 5 h, respectively. Figure 8 shows a snapshot of the three observed mixtures (see Supporting Methods 1 for more details about their construction). As in the binary systems, the three systems show cations and water molecules within the channel (Figure S13 shows the RMSF of each polymer).

FIGURE 8.

Snapshots of TAP:CyCo4:CyCo6 supramolecular polymers after 100 ns. Some sodium cations were omitted for clarity.

The HBs analysis of Table 2 indicates the inner HB system is less stable than the outer one as seen by the lower HB%, while the middle HBs still have 100% of occupancy. Furthermore, the HB% of the N − H···N interactions between amine groups in the channel increases as the ratio CyCo6/CyCo4 increases.

TABLE 2.

Weighted average of occupancy percentages (HB%) and average number of HBs per picosecond (N°) after 100 ns for TAP:CyCo4:CyCo6 polymers.

Composition	In a		Mid b		Out c		N − H···N d
	HB%	N°	HB%	N°	HB%	N°	HB%	N°
1:0.80:0.20	69	93.6	100	119.6	81	108.4	16	8.5
1:0.50:0.50	73	97.0	100	119.4	82	108.8	29	7.6
1:0.33:0.67	70	95.0	100	119.5	73	109.4	31	8.3

^a Inner interactions: N_TAP − H∙∙∙O_Cy‐R1.

^b Middle interactions: N_Cy‐R1 − H∙∙∙N_TAP.

^c Outer interactions: N_TAP − H∙∙∙O_Cy‐R1.

^d Hydrogen bonds between amine groups in the channel.

The hypothesis regarding these mixtures stated that the CyCo6 cogelator frustrates linear interactions between the stacks [12]. However, our simulations reveal that the addition of CyCo6 resulted in the formation of enhanced hydrophobic interactions between the side chains. As illustrated in Figure 9, the RDF analysis indicates that the mixture with the longest lifetime and the highest proportion of CyCo6 (1:0.33:0.67) exhibits a second peak at ∼5 Å. This finding suggests that, throughout the trajectory, the polymer adopts a helical twist of 30°, thereby facilitating the formation of side chain interactions.

FIGURE 9.

(a) Molecular structures of CyCo4 and CyCo6 monomers. The highlighted N atoms (green) were used as a reference point to analyze the helical twist between the stacks. (b) Radial pair distribution functions between highlighted N atoms of TAP:CyCo4:CyCo6 polymers. Green and magenta lines correspond to the reference systems with perfect helical twist. The data were collected at the 100th nanosecond.

3.3. Rational Design Strategy

Overall, there are two main factors that destabilize these types of polymers. First, the penetration of sodium cations into the channel that weakens the inner HBs, and second, the absence of hydrophobic interactions between side chains. If these two factors could be avoided, the polymer's stability would be expected to be even greater. In a previous study [14], we have computationally demonstrated that a triaminotriazine (TAT) moiety functionalized with a succinate group, TATsa (see Figure 10a), exhibits superior self‐assembly capability compared to its TAP analogue (TAPAS). Our molecule, TATsa, has been engineered to facilitate the coordination of Na⁺ cations without compromising its HB donor‐acceptor functionality. Therefore, by combining this programmed information with that of the CyCo6 molecule (see also Figure 10a), we may be able to create a supramolecular assembly that can bypass the two destabilizing factors mentioned above.

FIGURE 10.

(a) Molecular structure of the TAT moiety modified with a succinic acid group (sa), and the CyCo6 molecule obtained by Hud et al. [12] (b) Snapshot of the hydrogen‐bonded rosette within the polymer. (c) Snapshot of the designed supramolecular polymer after 100 ns with sticks representation (left, no sodium cation enters the inner channel) and van der Waals representation (right).

Figure 10bc show a snapshot of the TATsa:CyCo6 rosette and polymer, respectively, after 100 ns of simulation. As predicted, no Na⁺ enters the polymer channel and they get stuck near the cavity exposed to the solvent (see Figure 10c, left). The rest of the cations get coordinated by the TATsa molecule along the polymer, as shown in the snapshot on the right of Figure 10c. Consequently, the inner HBs are not perturbed and they become much stronger. As shown in Table 3, the HB% increases from ∼70% in previous polymers (TAP:Cy‐R1) to 93% in the TATsa:CyCo6 polymer (see also Figure S14).

TABLE 3.

Weighted average of occupancy percentages (HB%) and average number of HBs per picosecond (N°) after 100 ns for TATsa:CyCo6 polymers.

Polymer	In a		Mid b		Out c
	HB%	N°	HB%	N°	HB%	N°
TATsa:CyCo6	93	108.6	98	116.0	78	98.1

^a Inner interactions: N_TATsa − H∙∙∙O_CyCo6.

^b Middle interactions: N_CyCo6 − H∙∙∙N_TATsa.

^c Outer interactions: N_TATsa − H∙∙∙O_CyCo6.

When analyzing the helical twist by inspection of the RDF analysis (see Figure S15), the TATsa:CyCo6 polymer shows a peak at 4.45 Å and a very small shoulder at 4.85 Å. This result indicates the helical twist is shifted toward a smaller angle than 30°. This nearly ideal twist, which can be appreciated in Figure 10c (right side), promotes more hydrophobic side chain interactions. The cations are also coordinated by TATsa molecules (green molecules in Figure 10c) along the polymer surface. Summarizing, after carefully investigating the physicochemical factors that promote the precipitation of a set of supramolecular hydrogels, we were able to use these factors to design a new system that may show an enhanced stability.

4. Conclusion

Our MD and quantum chemical analyses provide a mechanistic explanation for why some supramolecular hydrogels precipitate over time, such as those obtained by the supramolecular polymerization of TAP with CyCo6 or CyCo4. In this study, we demonstrate that two primary factors are involved. First, sodium cations and hydration water molecules can get into the polymer channel through the cavities that are in contact with the solvent. This penetration can occur via two mechanisms. On the one hand, cations can enter the cavity through the ends of the polymer. However, given that experimental observations confirm these polymers are sufficiently long, the low frequency of end‐sites ensures this mechanism has a negligible impact on their stability. The second − and more dominant − mechanism involves penetration of cations through cavities exposed to the solvent due to the sliding of adjacent rosettes. Once sodium cations and water molecules migrate inside the channel, they establish multiple coordination and hydrogen bonding interactions with the carbonyl and amine groups of the channel. This process disrupts the inner HBs of the rosettes and promotes the polymer dissociation. The second instability factor is the absence of hydrophobic side chain interactions. These interactions are also responsible for the helical twist of the polymer. The predicted helical twists (30°) of TAP:CyCo6 and TAP:Cy4MeCo6 polymers were in excellent agreement with the experimental results.

The second experimental observation that we were able to explain computationally in this work is the role of CyCo6 as a cogelator. When this monomer is added to the TAP:CyCo4 mixture, the CyCo6 side chains can interact with each other as the ratio CyCo6/CyCo4 increases. Consequently, when these interactions are present, the lifetime of the polymer increases significantly.

Finally, our MD experiments were able to capture some of the most important structural features of these polymers in solution, as well as the main driving forces behind the lifetime and precipitation of the hydrogels. Therefore, a comprehensive physicochemical understanding of the involved phenomena was essential to redesign the available systems in silico with the required properties. Hence, the new system was able to avoid the entry of sodium cations into the polymer and improve the interactions between side chains. Our findings thus provide valuable information about the design principles for supramolecular materials in noncovalent synthesis.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting file: chem71149‐sup‐0001‐SuppMat.pdf

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.