Metal‐Free Active Template: A Straightforward Route to 4‐Aminopyridinium‐Based Rotaxane Molecular Shuttles

ABSTRACT

We report on the metal‐free active template synthesis of [2]rotaxanes through nucleophilic aromatic substitution between an amine and halogenopyridinium derivatives in the presence of a crown ether as the reaction promoter. The influence of the leaving group and counter‐ion of the pyridinium moiety was studied. Optimal conditions gave conversion of macrocycle to rotaxane up to 90% with rotaxane:axle ratio as great as 96:4. The ability of the 24‐crown‐8 and dibenzo‐24‐crown‐8‐containing [2]rotaxanes to behave as molecular shuttles was also considered. Mesomeric effects in the original 4‐aminopyridinium moiety revealed two discrete, though conjugated, sites of interaction for crown ethers, which allowed for small‐amplitude shuttling of the macrocycle. Unexpectedly, crown ether macrocycles mainly interacted around the amino site of the 4‐aminopyridinium conjugated system, highlighting the more preponderant iminium mesomeric form. Sole deprotonation of the amino group led to a neutral ylide‐kind dipolar [2]rotaxane, in which the crown ether freely shuttled along the entire axle. However, deprotonation‐then‐carbamoylation of the amino group triggered the small‐amplitude shuttling of the macrocycle around the sole positively charged pyridinium site of the 4‐aminopyridinium moiety.

Leigh's metal‐free active template‐based S_NAr of an amine on halogenopyridiniums, in the presence of a crown ether, successfully afforded 4‐aminopyridinium‐containing rotaxanes. Further molecular machinery was carried out by deprotonation‐then‐carbamoylation of the conjugated amino moiety. Carbamoylation hampered the amino site and revealed the pyridinium station by decreasing mesomeric delocalization of the amino moiety on the pyridinium site.

1. Introduction

Mechanically interlocked molecules (MIMs) [1]—such as rotaxanes [2], catenanes [3, 4, 5], or topologically non‐trivial molecules like molecular knots [6, 7, 8]—occupy an important place in organic and supramolecular chemistry. Molecular knot architecture is mathematically defined as a closed loop that contains entanglements. It cannot be untied unless the loop is open through breaking a covalent bond. In [n]rotaxanes and catenanes (n > 1), the n embedded components are interconnected not by conventional covalent bonds, but rather by their topological entanglement through mechanical bonds [9]. This characteristic endows MIMs with properties that have enabled the development of molecular switches [10, 11], catalysts [12, 13, 14, 15], electronics [16], and materials [17, 18] with unique physical and chemical properties. [2]Rotaxane architecture consists of a macrocycle threaded onto an axle: the presence of a bulky group at each extremity of the axle—commonly named stoppers—preserves the mechanical bond by sterically preventing the macrocycle from threading off the axle. It was in the early eighties that a passive template‐based strategy [19, 20, 21] was used for the first time to synthesize efficiently rotaxanes and catenanes. The method relies on noncovalent interactions to preorganize precursors in such a suited spatial supramolecular arrangement that the formation of the mechanically interlocked product becomes effective. In contrast, the active template (AT) synthesis method, first proposed by Leigh in 2006 [22, 23], represents a straightforward one‐step way synthetic route to synthesize rotaxanes, without the need for formal complementary recognition moieties between axle and the macrocycle precursors. Through metal active template synthesis, a metal predisposes the reactants and catalyzes their reaction that yields the interlocked compounds [22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43]. In the more recently reported metal‐free active template [44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54], the macrocycle promotes the interlocking reaction by stabilizing the transition state of the rate‐limiting step to the detriment of the axle formation [55]. Metal‐free AT synthesis of rotaxanes has been reported during S_N2 reactions, Michael additions, acylation reactions, or nucleophilic aromatic substitution (S_NAr) reactions on benzene derivatives. Despite these advances, the range of reactants and bond‐forming reactions compatible with AT synthesis remains limited. Only a few examples of metal‐free AT S_NAr‐based rotaxane synthesis have been reported to date, and they only concern benzene derivative reactants. In this regard, we explore herein the opportunity to exploit the pyridinium group as a reactant for AT synthesis of rotaxanes. Though monopyridinium units have been reported as secondary stations for dibenzo‐24‐crown‐8 (DB24C8) [56, 57, 58, 59], they have been largely excluded from the synthesis of interlocked molecules through passive methodology because of their too‐poor binding affinity for crown ethers. As an electron‐deficient aromatic ring, we thought pyridinium should be a good candidate for a metal‐free AT strategy in an S_NAr reaction, if combined with a suitable nucleophile amine. In the presence of a crown ether (24C8 or DB24C8), we thus envisioned the reaction between bis(trifluoromethyl)benzylamine and halogenopyridinium reactants, both bearing bulky benzyl units to ensure the mechanical bond in the targeted [2]rotaxanes (Figure 1, Scheme 1). We particularly focused on the influence of the nature of the halogen atom carried by the pyridinium ring and of the counteranion on the reactivity and selectivity toward rotaxane formation (Table 1). In a second part, we assessed the ability of the 4‐aminopyridinium‐containing [2]rotaxanes to behave as molecular shuttles (Figure 1, Scheme 2) by describing the distinct co‐conformational states obtained upon both deprotonation and carbamoylation of the amine moiety. Due to conjugation, it is noteworthy that the pH‐responsive system belonging to the [2]rotaxanes reported here (Figure 1b) is uncommon when referring to the popularly encountered ammonium/amine one (Figure 1a). Indeed, conjugation between the amine's nitrogen lone pair and π electrons of pyridinium allows for an invertible amine‐to‐amide acid‐base reaction instead of the routine amine‐to‐ammonium transformation. At the positively charged state—that is, aminopyridinium—the conjugation of the N₁₁ lone pair of electrons belonging to the amine group with the pyridinium moiety allowed for delocalization of the positive charge on nitrogen atom N₁₁. This revealed an iminium mesomeric form, which acted as the best site of interaction for the considered crown ethers. Deprotonation of the amine function into neutral ylide‐kind pyridinium amide rotaxanes resulted in the decrease of the strength of interactions between the crown ether and the axle, so that the crown ether shuttled along the entire axle. However, deprotonation and then N‐carbamoylation of the amine concealed the amine site moiety. It also favored the pyridinium mesomeric structure because the introduction of the electron‐withdrawing group Boc decreased the resonance between the lone pair of electrons of N₁₁ and the pyridinium unit. Hence, the small‐amplitude shuttling of the crown ether around the sole pyridinium site occurred.

FIGURE 1.

(a) Previously reported large‐amplitude pyridinium‐ and ammonium‐containing two‐station [2]rotaxane molecular shuttle obtained via a passive template strategy [56, 57, 58, 59]; (b) current work: original targeted 4‐aminopyridinium‐containing [2]rotaxanes synthesized through an S_NAr‐based metal‐free active template, then molecular shuttling of the macrocycle upon deprotonation and carbamoylation.

SCHEME 1.

Effect of the nature of the halogen atom X, the counteranion Y⁻, temperature, and solvent on the synthesis of pyridinium‐based [2]rotaxanes 3H_24C8‐Y and 3H_DB24C8‐Y using the metal‐free active template strategy.

TABLE 1.

Conversion ratio of macrocycles, isolated yields, and the 3H:3uH ratios for the synthesis of pyridinium‐based [2]rotaxanes 3H from pyridinium derivatives 1 ([1] = 0.142 M, 1 equiv.) and amine 2 (1 equiv.) using the active template strategy, under variations in the solvent, temperature, nature of the halogen atom X, counteranion Y⁻, and amount of Et₃N.

			Equiv. of
Entry	X	Y	1	2	CE c	Et3N	DBU	Solvent	T (K)	Time	Conv a of CE (%)	Ratio b 3HCE:3uH	Isol yield for 3HCE (%)
1	Cl	Br	1	1	1 d	2	—	Toluene	298	1 day	25	30:70	21
2	Cl	Br	1	1	2 d	2	—	Toluene	298	1 day	29	33:67	—
3	I	Br	1	1	1 d	2	—	Toluene	298	3 days	30	32:68	—
4	I	Br	1	1	1 d	2	—	Toluene	273	2 days	17	47:53	—
5	I	PF6	1	1	1 d	2	—	Toluene	273	7 h	48	92:8	—
6	I	PF6	1	1	1 d	30	—	Toluene	273	8 h	66	94:6	—
7	I	PF6	1	1	1 d	30	—	Toluene	273	3 days	90	90:10	85
8	I	PF6	1	1	1 d	30	—	Toluene	298	3 days	77	84:16	—
9	I	PF6	1	1	1 d	30	—	DCM	273	8 h	62	93:7	—
10	I	PF6	1	1	1 d	30	—	DCM	298	2 h	57	95:5	—
11	I	PF6	1	1	1 d	30	—	DCM	298	16 h	61	95:5	—
12	I	PF6	1	1	1 d	—	2	Toluene	273	3 days	62	74:26	—
13	I	PF6	1	1	1 e	30	—	DCM	273	36 h	70	96:4	—
14	I	PF6	1	1	1 e	30	—	Toluene	273	36 h	59	86:14	—
15	I	PF6	1	1	1 e	30	—	Toluene	273	3 days	73	87:13	—
16	I	PF6	1	1	1 e	30	—	Toluene	273	5 days	74	85:15	71

^a Measured on ¹H NMR spectroscopy through integration of the H_E belonging to CE and 3H_CE (see Figures S4.1–S4.3).

^b Measured on ¹H NMR spectrum by integrating signals of H₇, H₁₂ or H_9a for rotaxane 3 and thread 3uH.

^c CE means crown ether.

^d CE = 24C8.

^e CE = DB24C8.

SCHEME 2.

Deprotonation, carbamoylation, and then decarbamoylation of the amine moiety of rotaxanes 3H_CE‐PF₆ and free axle 3uH‐PF₆ .

2. Results and Discussion

2.1. Synthesis of 24C8 and DB24C8‐Containing [2]Rotaxanes 3H_CE‐PF₆ via Metal‐Free Active Template‐Based S_NAr

Before starting our investigation, we made sure that each precursor of the axle could not yield pseudorotaxane via a passive template pathway when mixed with the considered crown ethers (see Supporting Information, Section 8). Then, in a first attempt, a mixture in toluene of 4‐chloropyridinium bromide derivative 1a‐Br (1 equiv., [1] = 0.142 M), 3,5‐bis(trifluoromethyl)benzylamine (1 equiv.), 24C8 (1 equiv.), and triethylamine (2 equiv.) was stirred at room temperature, and the reaction was monitored by ¹H NMR (entry 1, Table 1). After 1 day, ¹H NMR spectrum of the crude reaction mixture indicated the formation of rotaxane 3H_24C8 according to a low conversion of 24C8 into rotaxane (25%) and with an encouraging but unsatisfactory ratio of 3H_24C8 :3uH 30:70. In the same experimental conditions, increasing the quantity of 24C8 did not provide significantly better results (Table 1, entry 2). In order to improve the selectivity toward rotaxane product 3H_24C8 , we decided to decrease the reactivity of the starting pyridinium derivative. Hence, replacing chlorine with the less electron‐withdrawing iodine atom was first considered, with the aim of slowing down the nucleophilic addition of the amine during the rate‐limiting step of the S_NAr. Though conversion of the pyridinium derivative 1b‐Br proved to be slower (3 days to reach 96% conversion of reactant 1b‐Br), no significant improvement was noticed in the ratio 3H_24C8 :3uH, nor in the conversion of 24C8 into rotaxane (Table 1, entry 3). Carrying out the reaction from 1b‐Br at a lower temperature (273 K instead of 298 K) resulted in a slightly better ratio of 3H_24C8 :3uH (47:53), although it was still insufficient (Table 1, entry 4). In order to further improve the selectivity toward rotaxane species, changing the pyridinium counteranion for the hexafluorophosphate anion was then envisioned. We expected its higher softness related to the hard and soft acids and bases theory should provide a loose ion pair that would result in a less electron‐deficient pyridinium derivative, therefore decreasing the pyridinium reactivity toward S_NAr. Moreover, it should increase the poor solubility of the pyridinium reactant in toluene. Finally, compared to ammonium bromide, the loose ion pair, such as ammonium hexafluorophosphate, was known to behave as a better template for crown ether in a passive template strategy. Therefore, with this loose ion pair, crown ether should stabilize better the transition state of the rate‐limiting step during the considered metal‐free active template‐based S_NAr, thus favoring rotaxane products over free axles. Pleasingly, this variation caused a tremendous change in the selectivity of the reaction, which yielded mainly rotaxane 3H_24C8 with only traces of uncomplexed thread 3uH after 7 h at 273 K (Table 1, entry 5). By analyzing the ¹H NMR of the reaction mixture, we pointed out at this time the remaining significant quantity (approximately 50%) of the starting pyridinium derivative. We also noticed that the remaining amine 2 was mainly under the protonated form in the reaction mixture due to the release of H‐I during the elimination second step of the mechanism. The sequential increase of the amount of triethylamine to 30 equiv., then of the reaction time to 3 days, allowed for almost full conversion of both the pyridinium reactant and 24C8, with a ratio of 3H_24C8 :3uH better than 90:10 (Table 1, entries 6–7). Carrying out the reaction at room temperature did not improve the conversion yield to 3H_24C8 nor the ratio 3H_24C8 :3uH (Table 1, entry 8). Instead, 5% of the side‐product 4‐pyridone 4 was detected along with 10% of the unexpected compound 5 (Scheme 1 and Section 7 of Supporting Informaton). Pyridone 4 and aminopyridinium 5 resulted from side reactions of 1b‐PF₆ , in the presence of triethylamine and residual water. An additional experiment was carried out to confirm the degradation of 1b‐PF₆ into compounds 4 and 5 by stirring a solution of 1b‐PF₆ in dichloromethane at room temperature in the presence of triethylamine and residual water. Noteworthy, the addition of water to 1b‐PF₆ did not lead to pyridone 4 in the absence of triethylamine. To confirm that side products were the unexpected pyridone 4 and aminopyridinium 5, both compounds were separately efficiently synthesized from 1b‐PF₆ using other ways, by adding potassium hydroxide in THF or by adding diethylamine, respectively (see section 7 of the SI). A suggested mechanism for the formation of 5 via rearomatization‐driven dealkylation of a triethylammonium moiety is provided in the Supporting Information (Section 7.4).

Besides, during active‐template synthesis of rotaxanes, it was noticed that the reaction medium was not homogeneous in toluene, partially due to the weak solubility of the pyridinium. Since this could explain the slow rate of the reaction, hydrogen bond‐promoting solvent dichloromethane was considered to remedy this issue. No precipitate was observed in this new solvent during the reaction, and 87% conversion of the pyridinium derivative 1b‐PF₆ was noticed after 8 h at 273 K, still with an excellent ratio of 93% for rotaxane 3H_24C8 with respect to 3uH (Table 1, entry 9). Interestingly, similar excellent results in terms of the rotaxane‐to‐thread ratio (3H_24C8 :3uH) were obtained in dichloromethane at room temperature after only 2 h (Table 1, entry 10). Nevertheless, despite the better solubility of the reactants in dichloromethane, full conversion of the pyridinium reactant 1b‐PF₆ after 16 h did not afford a better conversion yield of the 24C8 into rotaxane 3H_24C8 than when the reaction was led in toluene until completion (Table 1, entries 11 and 7, respectively). Indeed, in the better solubilizing solvent dichloromethane, side reactions of triethylamine and residual water on pyridinium reactant 1b‐PF₆ seem to be slightly more competitive than in toluene, providing the two identified side products 4 and 5 (Scheme 1), with conversion yields of 6%–8% and 8%–14%, respectively (Table 1, entries 9–11). Using 1,8‐diazabicyclo[5.4.0]undec‐7‐ene (DBU) instead of triethylamine allowed to avoid the formation of the side‐compound 5, although it did not improve the conversion yield of 24C8 into rotaxane nor the selectivity toward rotaxane with respect to the axle (Table 1, entry 12). Replacing 24C8 by DB24C8 did not affect significantly the efficiency of the metal‐free active template reaction. For the same reasons as discussed above, a better conversion of DB24C8 into rotaxane 3H_DB24C8 was obtained in toluene with triethylamine as the base, along with similar high selectivity for rotaxane synthesis with respect to the uncomplexed thread, whatever the solvent used (3H_DB24C8‐Y:3uH > 85:15) (Table 1, entries 13–16). In comparison to 24C8, the reaction appears slightly slower with the DB24C8 (still 8% of the starting pyridinium reactant 1b‐PF₆ remained unreacted after 3 d in toluene at 273 K) (Table 1, entries 7 and 15). Besides, contrary to rotaxane 3H_24C8 , rotaxane 3H_DB24C8 quantitatively precipitated in toluene. For the best tries (Table 1, entries 7 and 16), each crude reaction mixture was treated in order to isolate each of the 24C8 and DB24C8‐containing [2]rotaxanes. Since iodide anion was released during S_NAr, treatment with ammonium hexafluorophosphate was carried out to make sure that only one pair of ions was obtained. Each rotaxane (3H_24C8‐PF₆ or 3H_DB24C8‐PF₆ ) was then purified by size exclusion chromatography and obtained in 85% (entry 7), or 71% (entry 16) yield, respectively.

2.2. ¹H NMR and X‐Ray Evidences of the Interlocked Molecular Architecture of [2]Rotaxanes 3H_CE‐PF₆

The ¹H NMR spectra comparison between free molecular axle 3uH‐PF₆ , rotaxanes 3H_24C8‐PF₆ or 3H_DB24C8‐PF₆ , and free crown ethers 24C8 or DB24C8 gave us insights about the interlocked structure and main localization of the crown ether along the encircled axle in each rotaxane (Figure 2).

FIGURE 2.

¹H NMR (400 MHz, CDCl₃, 298 K) spectra comparison between (a) free 24C8; (b) rotaxane 3H_24C8‐PF₆ ; (c) uncomplexed thread 3uH‐PF₆ ; (d) rotaxane 3H_DB24C8‐PF₆ ; (e) free DB24C8. The numbering and lettering correspond to the proton assignments indicated in Scheme 1.

With respect to the free 24C8, the H_E hydrogen atoms become inequivalent and are therefore split in rotaxane 3H_24C8‐PF₆ , because they face the two distinct extremities of the unsymmetric encircled thread (Figure 2a,b). They are also slightly shielded in 3H_24C8‐PF₆ (Δδ = −0.28 and −0.38 ppm, respectively) because they experience the shielding effect of the aromatic bis(trifluoromethyl)phenyl ring. In addition, comparing the interlocked compound with its uncomplexed thread 3uH‐PF₆ allowed to determine the main localization of the 24C8 around the “right part” of the threaded axle (Figure 2b,c). We could have initially thought that crown ether would prefer to interact with the positively charged pyridinium moiety rather than with the uncharged benzyl site. Actually, the resonance of the 4‐aminopyridinium moiety, through the mesomeric electron‐donating amino group, which tends to delocalize the positive charge until the nitrogen atom N₁₁, combined with the electron‐withdrawing trifluoromethyl moieties, accounts for the localization of the 24C8 around the more hydrogen‐bond‐donating atoms H_9a, H₁₂ and H₁₄. Indeed, hydrogen atoms H_9a, H₁₂ and H₁₄ are shifted downfield in rotaxane 3H_24C8‐PF₆ (Δδ = 0.55, 0.23 and 0.56 ppm, respectively) because they interact through hydrogen bonding with the oxygen atoms of 24C8. The mesomeric effect (+M) of the amino group, which confers a π character to the C₁₀‐N₁₁ bond, also explains the splitting of ¹H NMR signal of H₉ on the one hand and that of H₈, on the other hand. Its quantification, which was evaluated by determining the kinetic data relative to the rotation of this bond, is reported later in the article. In the meantime, H_9b is shifted upfield (Δδ = −0.46 ppm) because of the ion‐dipole interaction between the positive charge and the lone pairs of electrons belonging to oxygen atoms of the 24C8. Finally, no significant chemical shift variation was noticed for hydrogen atoms H₁, H₄, H₅, H₇ and H₈ between rotaxane and thread, which corroborates the main localization of the 24C8 around the “right part” of the axle. A similar trend in the downfield chemical shift displacement for H₁₂ and H₁₄ was observed in rotaxane 3H_DB24C8‐PF₆ with respect to the thread 3uH‐PF₆ (Δδ = 0.29 and 0.61 ppm, respectively) (Figure 2c,d), confirming similar main localization of the DB24C8 around the “right part” of the axle. However, contrary to 3H_24C8‐PF₆ , the hydrogen atoms belonging to the pyridinium region H₈, H_9b, and neighboring atoms H₇ are tremendously shielded in the rotaxane 3H_DB24C8‐PF₆ (Δδ = −0.65, −0.83 and −0.45 ppm, respectively), suggesting additional donor‐acceptor π–π interactions between the electron‐rich aromatic rings of the DB24C8 and the electron‐deficient pyridinium ring. Besides, the comparison between free DB24C8 and rotaxane 3H_DB24C8‐PF₆ confirmed the interlocked nature of the rotaxane and the predominant localization of the macrocycle along the thread (Figure 2d,e). Indeed, H_A‐E appears all split in the rotaxane because of the unsymmetric thread. Moreover, H_A‐B are slightly shielded (Δδ = −0.22, −0.28 ppm, respectively) due to stacking, while H_E,E’ are importantly shielded in the rotaxane (Δδ = −0.58 ppm)—and in a lesser manner H_C,C’ and H_D,D’ with respective Δδ = −0.14 and −0.23 ppm—because they experience the shielding region of the aromatic bis(trifluoromethyl)phenyl stoppering unit. 2D ROESY ¹H NMR experiments were also carried out for each rotaxane 3H_24C8‐PF₆ or 3H_DB24C8‐PF₆ (see Figure 4a,d). They confirmed the accurate main localization of each crown ether around the “right part” of the thread. Indeed, cross‐peaks were observed between the hydrogen atoms H_E or H_C‐E of the respective 24C8 or DB24C8 and the hydrogen atoms H_9,11,12,14 belonging to the “right part” of the encircled axle. Moreover, in 3H_DB24C8‐PF₆ , cross‐peaks between H_A‐B of the DB24C8 and H_4,5,7 corroborated the π–π stacking between the pyridinium ring and aromatic rings of the DB24C8.

FIGURE 4.

2D ¹H NMR ROESY (600 MHz, CDCl₃, 298 K) excerpts indicating correlation peaks between hydrogen atoms of crown ethers and axle in rotaxanes: (a) 3H_24C8‐PF₆ ; (b) 3_24C8 ; (c) 3Boc_24C8‐PF₆ ; (d) 3H_DB24C8‐PF₆ ; (e) 3_DB24C8 ; (f) 3Boc_DB24C8‐PF₆ . The numbering and lettering correspond to the proton assignments indicated in Scheme 2.

Pleasingly, the solid‐state structures of rotaxane 3H_DB24C8‐PF₆ could be determined by x‐ray diffraction (CCDC 2533606; space group P 2₁/c with four rotaxane molecules, PF₆ counteranions, and 6 molecules of toluene in the unit cell) on crystals obtained from slow evaporation of a dichloromethane‐toluene solution. X‐ray crystal structure confirmed the co‐conformations deduced from ¹H NMR studies (Figure 3a–c). As it appeared to be in solution, in the solid state, the DB24C8 is localized around the N‐bis(trifluoromethyl)benzyliminium site and hydrogen bonding interactions between oxygen atoms of the DB24C8 and H₁₂ are noticed. In the meantime, π–π stacking between the axle pyridinium ring and the two DB24C8 phenyl rings is noted. The slightly imperfect superposition of the aromatic rings (Figure 3b) may explain the ¹H NMR shielding of only one of the two hydrogen H₉ in 3H_DB24C8‐PF₆ (Figure 2c,d).

FIGURE 3.

X‐ray crystal structure representations of 3H_DB24C8‐PF₆ , 3Boc_DB24C8‐PF₆ and 7H_24C8‐PF₆ . For clarity, only hydrogen atoms of the axle, which may be affected by the presence of the DB24C8 are mentioned in 3H_DB24C8‐PF₆ and 3Boc_DB24C8‐PF₆ (see 1D‐¹H NMR of Figure 2). Color for carbon atoms of each interlocked unit corresponds to that indicated in Scheme 1, while oxygen, nitrogen, hydrogen, and fluorine atoms are indicated in red, blue, white, and yellow, respectively. Molecules of solvent (toluene for 3H_DB24C8‐PF₆ and dichloromethane for 3Boc_DB24C8‐PF₆ ) and counteranion hexafluorophosphate have been removed. (a, b) Stick model representations of 3H_DB24C8‐PF₆ highlighting hydrogen bonds involving H₁₂ and π–π stacking between the aromatic rings of the DB24C8 and the pyridinium ring belonging to the axle. (c) Space filling representation of 3H_DB24C8‐PF₆ : all atoms are represented. (d, e) Stick model representations of 3Boc_DB24C8‐PF₆ highlighting the position of the DB24C8 around the pyridinium site through hydrogen bonds involving H₇ and H₈. (f) Space filling representation of 3Boc_DB24C8‐PF₆ : all atoms are represented. (g, h) Stick model representations of 7H_24C8‐PF₆ . (i) Space‐filling representation of 7H_24C8‐PF₆ : all atoms are represented.

2.3. Shuttling of the Macrocycle Along the Thread and ¹H NMR Evidences of the Molecular Machinery

With the aim of controlling the shuttling of the macrocycle along the thread, we envisioned the successive deprotonation and carbamoylation of the NH₁₁ group of each rotaxane 3H_CE‐PF₆ and studied the localization of the macrocycle in each rotaxane product through ROESY ¹H NMR experiments (Scheme 2). Noteworthy, all rotaxanes described thereafter appeared kinetically stable at room temperature when stored dry, as well as in the solvents used during synthesis, purification, characterization, and crystallization (dichloromethane, chloroform, and diethylether). Use of highly dissociating solvents such as dimethyl sulfoxide and higher temperatures were not used in order to prevent any possibility of dethreading through the tert‐butyl extremity [60].

Deprotonation was carried out using the basic phosphazene‐based BEMP resin, which afforded quantitatively the original neutral [2]rotaxanes 3_24C8 and 3_DB24C8 . Due to their high ability to protonate back to initial reactants, neutral rotaxanes 3_CE and free axle 3u were characterized by ¹H NMR spectroscopy in the presence of the BEMP resin. Noteworthy, deprotonated rotaxanes and axles may be drawn under two main resonance structures: one of them consists of the uncharged imine mesomeric form, while the other is the neutral ylide‐kind pyridinium amide form. A resonance hybrid indicates the electrophilic sites that are prone to interact through hydrogen bonding with the oxygen atoms of the crown ether. After deprotonation, 2D ROESY ¹H NMR of both rotaxanes 3_24C8 and 3_DB24C8 indicated that both macrocycles shuttle along the entire encircled threads. Indeed, spatial proximity between hydrogen atoms of the crown ether macrocycle with hydrogen atoms of both the “left” (1D ¹H NMR signals of H_4‐9 in blue) and “right” (H_12‐16 in green) parts of the axle could be evidenced in Figure 4b,e. The single set of ¹H NMR signals observed here suggested a fast translating process along the treaded axle on the NMR time scale. Freezing the shuttling of the macrocycle between the two sites of interactions was attempted at low temperature in order to distinguish better the main co‐conformers. Unfortunately, we did not see any splitting of the signals arising from both co‐conformational isomers by VT ¹H NMR. Though, a preference for the initial bis(trifluoromethyl)benzylamine “green right part” was noticed if one refers to the intensities of ROESY correlation peaks. This behavior of the crown ethers along the thread meant that the uncharged imine mesomeric form is preponderant over the pyridinium amide one. Indeed, if this were not the case, we would have expected the macrocycle to solely reside around the pyridinium site, as it would have behaved with a non‐conjugated amine‐pyridinium system.

In order to obtain a predominant localization of the crown ether around the pyridinium site of interaction, carbamoylation of each rotaxane 3_CE was carried out in chloroform using Boc₂O in the presence of the BEMP resin (Scheme 2). Purification using size exclusion chromatography provided pure rotaxanes 3Boc_24C8‐PF₆ and 3Boc_DB24C8‐PF₆ in the respective 61% and 59% overall yield, starting from protonated rotaxanes 3H_24C8‐PF₆ and 3H_DB24C8‐PF₆ . Due to the concealing of the bis(trifluoromethyl)benzylamino site, the macrocycle shuttled and mainly resided around the sole pyridinium site. Correlation peaks observed through ROESY experiments (Figure 4c,f) showed unambiguous localization of each crown ether around the “left blue” part of the surrounded axle. In this case, contrary to 3H_CE‐PF₆ , the introduction of the mesomeric electron withdrawal group Boc makes the delocalization of the lone pair of electrons belonging to exocyclic nitrogen atom N₁₁ with the pyridinium unlikely. Thus, the pyridinium site becomes the best site of interaction for the crown ethers. The possible delocalization of the positive charge between nitrogen atom of the pyridinium moiety and the exocyclic nitrogen atom N₁₁ could be studied easily by ¹H NMR. In rotaxane 3Boc_CE‐PF₆ and the related uncomplexed axle 3uBoc, hydrogen atoms H₈ (as H₉) appear as one single set of ¹H NMR signals. Their chemical equivalence then indicates a fast rotation of the C₁₀‐N₁₁ bond on the NMR time scale, which confirms that delocalization of the lone pair of electrons of the exocyclic nitrogen atom N₁₁ with the pyridinium moiety is unlikely. On the contrary, in rotaxanes 3H_CE‐PF₆ and 3_CE , as well as in their uncomplexed axles 3uH‐PF₆ and 3u, ¹H NMR signals of pyridinium hydrogen atoms H₉ (as H₈) are split into H_9a and H_9b at room temperature. This non‐equivalence of hydrogen atoms is here the result of the delocalization of the lone pair of electrons of N₁₁ to the pyridinium moiety, which gives a partial double bond character to the C₁₀‐N₁₁ bond, therefore inducing the slow rotation of the C₁₀‐N₁₁ bond on the NMR time scale. This observation correlates with the localization of the macrocycle around the iminium or imine site in the respective 3H_CE‐PF₆ and 3_CE .

Determination of kinetic data relative to the rotation of the C₁₀‐N₁₁ bond was achieved and gave corroborating insights into the localization of the crown ether around either the pyridinium or the amine site. For [2]rotaxanes 3H_CE‐PF₆ , 3_CE and analogous free axles, EXSY experiments allowed to estimate the exchange rate constants k_ex between H_9a and H_9b, then the related free enthalpies of activation ΔG ^‡ thanks to the Eyring equation (Table 2) (see Supporting Information, Section 5) [61].

TABLE 2.

Kinetic data relative to the exchange between non‐equivalent hydrogen atoms H_8a and H_8b or H_9a and H_9b (for free axles 3u, rotaxanes 3_CE and protonated and carbamoylated derivatives), and H_4a and H_4b (for rotaxane 7H_24C8‐PF₆ ) calculated from the 2D EXSY spectra (298 K, 600 MHz, mixing time τ _m = 0.3 s, relaxation delay t ₁ = 1.5 s).

	Free axles		Rotaxanes
Compound	3uH‐PF6 c	3u d	3H24C8‐PF6 d	3HDB24C8‐PF6 d	324C8 d	3DB24C8 d	7H24C8‐PF6 d
kex (s−1) a	9.11	0.10	2.79	1.88	2.49	0.11	4.68
ΔG ‡ (kJ·mol−1) b	67.5	78.8	70.4	71.4	70.7	78.4	69.2

^a

$k_{ex}=\frac{1}{{\tau }_m}ln[\frac{r+1}{r-1}],r=\frac{I_{\mathrm{AA}}+I_{\mathrm{BB}}}{I_{\mathrm{AB}}+I_{\mathrm{BA}}},$

^b $\mathrm{\Delta }{G}^{‡}=-RT\ln [\frac{k_{ex}h}{k_{B}T}]$, where τ _m as the mixing time; I denotes the volume intensities of diagonal and cross peaks in the 2D EXSY spectrum; R = 8.314 J·mol⁻¹·K⁻¹, h = 6.626×10⁻³⁴ J·s, and k _B = 1.381×10⁻²³ J·K⁻¹.

^c In CD₂Cl₂.

^d In CDCl₃.

As discussed, these values closely depend on the preponderance of the mesomeric forms of the 4‐aminopyridinium moiety (Figure 5). The exchange rate constant appeared circa 3 to 5 times lower in the positively charged rotaxanes 3H_CE‐PF₆ than in their respective uncomplexed thread 3uH‐PF₆ , demonstrating that delocalization of the positive charge from the pyridinium nitrogen to the exocyclic nitrogen iminium was better in rotaxanes 3H_CE‐PF₆ than in thread 3uH‐PF₆ . This result is consistent with hydrogen bonding‐based stabilization of the iminium form by the crown ether in the rotaxane species, the mesomeric iminium form thus becoming more preponderant in the rotaxane than in the free axle.

FIGURE 5.

Illustration of the correlation between mesomeric forms and kinetic rate of the C₁₀‐N₁₁ bond rotation.

After deprotonation, rotation of C₁₀‐N₁₁ bond tremendously decreased in free neutral axle 3u (91 times slower) compared to the free protonated axle 3uH‐PF₆ . As expected, this is the result of the higher mesomeric electron‐donating effect of the negatively charged amide group with respect to the initial amine moiety. Comparison of the rotation kinetic rate for C₁₀‐N₁₁ bond between neutral rotaxanes 3_CE and axle 3u proved to be dependent on the nature of the crown ether. Rotation of C₁₀‐N₁₁ bond was found faster (approximately 25 times) in the rotaxane 3_24C8 than in free axle 3u, while similar rates were calculated for free axle 3u and rotaxane 3_DB24C8 . In both rotaxanes, the partial presence of the crown ether around the pyridinium site participates to the acceleration of the bond rotation, by increasing the probability of the pyridinium mesomeric form—thence increasing the σ character of C₁₀‐N₁₁ bond—in the rotaxane species with respect to the free axle. However, in 3_DB24C8 , this effect seems to be offset by π–π stacking involving the DB24C8, which hampers the free rotation of the nitrogen heterocycle.

Single crystals of 3Boc_DB24C8‐PF₆ were grown through slow evaporation of a solvent mixture dichloromethane/diethylether and the structure was determined by x‐ray diffraction [62]. 3Boc_DB24C8‐PF₆ crystallizes in the monoclinic space group P2₁/n with 4 rotaxane molecules and 4 PF₆ ⁻ anions. The x‐ray crystal structure of 3Boc_DB24C8‐PF₆ is represented in Figure 3d–f. At the solid state, the main localization of the DB24C8 around the pyridinium site correlates with what was observed by NMR in solution. The crystal structure notably points out the hydrogen bonding pattern between pyridinium hydrogen atoms H₈ and neighboring atom H₇ with oxygen atoms of the DB24C8. Moreover, C₁₀‐N₁₁ bond lengths measured at the solid state are in accordance with what was discussed above about the pronounced π character of the C₁₀‐N₁₁ bond in 3H_DB24C8‐PF₆ (d = 1.33 Å) and the σ character of the same C₁₀‐N₁₁ bond in 3Boc_DB24C8‐PF₆ (d = 1.40 Å) [63].

Shuttling back of the macrocycle from the pyridinium to the iminium site (i.e., from 3Boc_CE‐PF₆ to 3H_CE‐PF₆ ) was achieved thanks to Boc removal under acidic conditions and subsequent anion exchange (Scheme 2).

2.4. Extension to the Synthesis of an Acridinium [2]Rotaxane

In order to exemplify the tolerance of the metal‐free active template‐based S_NAr reaction to structural variation of the pyridinium moiety, 9‐chloro‐10‐methylacridin‐10‐ium was considered as the aromatic reactant, this time featuring a more crowded reactive site (Scheme 3). In this case, the pyridinium derivative—a tricyclic acridinium aromatic moiety—was bulky enough to ensure the mechanical bond in the targeted [2]rotaxane product. Therefore, it was sufficient to connect a non‐hindering methyl group to the nitrogen atom of the acridine motif to provide the required electron‐poor acridinium. Compound 6‐OTf was freshly prepared through methylation of 9‐chlorochloroacridine using 1.1 equiv. of methyl trifluoromethylsulfonate in dichloromethane at 273 K. It led to compound 6‐OTf—a 9‐chloroacridinium trifluoromethylsulfonate loose ion pair—as a mixture with unsolved side‐compounds, which was directly used in the following rotaxane formation with or without prior triflate to hexafluorophosphate anion exchange. From the ¹H NMR spectra of the reaction mixtures, we calculated a ratio rotaxane 7H_24C8‐X versus thread 7uH‐X of 39:61 when the reaction was led from acridinium 6‐OTf and a better ratio of 54:46 from 6‐PF₆ , indicating once again the more suitable hexafluorophosphate anion for the metal‐free active template S_NAr. Comparing to the synthesis of rotaxane 3H_24C8‐PF₆ , we attributed the lesser selectivity for rotaxane 7H_24C8‐PF₆ to steric hindering brought by the substitutions of the pyridinium moiety around the reactive site, which hampered the good preorganization between chloroacridinium, 24C8 and bis(trifluoromethyl)benzylamine in the transition state of the rate‐limiting‐step. This may be seen by comparing x‐ray structures in Figure 3a,g: in 7H_24C8‐PF₆ , the conformational degree of freedom of the crown ether along the encircled axle is clearly reduced by the bulk of the acridinium moiety. Pure rotaxane 7H_24C8‐PF₆ was nevertheless isolated after precipitation in toluene, anion exchange, then BioBeads exclusion chromatography in a 40% yield from the commercially available bis(trifluoromethyl)benzylamine 2.

SCHEME 3.

Synthesis of the acridinium‐containing [2]rotaxane 7H_24C8‐PF₆ through metal‐free active template based S_NAr.

As for the previously described rotaxanes, kinetic data relative to the rotation of the C₆‐N₇ bond were calculated from the EXSY ¹H NMR experiment (see Table 2 and Supporting Information). The bond rotation was slightly quicker in the acridinium series than in the pyridinium one (rotaxanes 7H_24C8‐PF₆ and 3H_24C8‐PF₆ , respectively), certainly because the 24C8, which cannot surround the larger acridinium moiety, does not hamper its rotation as much as when it can surround the smaller pyridinium motif in 3H_24C8‐PF₆ . By comparison between rotaxane and thread, no splitting of the considered ¹H NMR signals was observed in the uncomplexed thread 7uH‐PF₆ , indicating a fast rotation of the C₆‐N₇ bond on the NMR time scale in the absence of the surrounding macrocycle, hence a lesser mesomeric delocalization of the lone pair of the nitrogen atom N₇. Crystals of rotaxane 7H_24C8‐PF₆ were obtained through slow evaporation of a solvent mixture of dichloromethane/toluene and x‐ray crystal structure was obtained (CCDC 2546575) (Figure 3g–i). It shows a pretty compact molecule, without any solvent molecules in the unit cell. The pronounced π character of the C₆‐N₇ bond (d = 1.32 Å) was confirmed, highlighting the more preponderant iminium mesomeric form in the rotaxane species. Another structural feature of rotaxane 7H_24C8‐PF₆ is the deviation from flatness of the central nitrogen‐containing cycle, displaying carbon C₁ and nitrogen N₇ in the cis position (Figure 3g). Values of 160.7° and 167.6° were noticed for respective angles (N₇‐C₆ ^…N₁) and (C₁‐N₁ ^…C₆) instead of the expected 180°.

3. Conclusion

In conclusion, we have reported the possibility of using Leigh's metal‐free active template strategy with two different crown‐ethers to efficiently produce 4‐aminopyridinium‐containing [2]rotaxanes through S_NAr reaction from a 4‐halogenopyridinium reactant. Hexafluorophosphate as the pyridinium counter‐ion proved to be the key success factor for high selective formation of rotaxane vs unthreaded axle species, while toluene as the solvent and low temperature afforded high yields by limiting side‐degradation of the pyridinium reactant. However, variation of the halide leaving group was not found to be essential. Small‐amplitude shuttling of the macrocycle between the two electrophilic sites of the original conjugated 4‐aminopyridinium moiety was also achieved through deprotonation‐then‐carbamoylation of the amine moiety. Interestingly, a correlation between delocalization of the positive charge through mesomeric effect in the conjugated moiety, position of the macrocycle, and rotation of the amine to pyridinium bond was pointed out. Noteworthy, the metal‐free active template strategy still proved its efficacy with the more crowded 9‐chloro‐10‐methylacridin‐10‐ium, though with a more modest selectivity for rotaxane. The possibility to use a wide range of pyridinium moieties would allow for straightforward access to many more sophisticated interlocked 4‐aminopyridinium‐based molecular shuttles, if obviously the steric hindrance of the pyridinium still allows for a good promoting preorganization between the crown ether and axle precursors.

Author Contributions

Ivaylo Stoyanov: investigation, methodology, formal analysis, data curation, resources, writing – review and editing. Arie van der Lee: formal analysis, writing – review and editing, investigation, data curation. Frédéric Coutrot: conceptualization, writing – original draft, investigation, funding acquisition, supervision, methodology, validation, writing – review and editing, formal analysis, data curation.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

The authors have cited additional references within the Supporting Information [64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75].

Supporting File: anie72605‐sup‐0002‐cif.zip.

Stoyanov I., van der Lee A., and Coutrot F., Angewandte Chemie International Edition. 65, no. 29 (2026): e8134791. 10.1002/anie.8134791

Data Availability Statement

The data that supports the findings of this study are available in the supplementary material of this article.