Molecular rotors have various potential applications in molecular devices1,2,3, medicine4,5,6, asymmetric catalysis7,8, and smart materials9,10,11, which were derived from their controllable rotations. For example, light-driven molecular motors could drill through cell membranes using their molecular-scale actuation, inducing necrosis and facilitating chemical species into live cells5. Feringa and co-workers developed an artificial muscle-like functional materials which was based on supramolecular assembly of photoresponsive molecular motors9. Therefore, it is of great significance to develop novel rotor systems with controllable manners. As an emerging class of macrocycles, pillar[n]arenes have attracted wide attentions due to their electron-rich cavities, and played important roles in the field of supramolecular chemistry12,13,14,15,16. Pillar[n]arenes have a unique planar chirality, in which two stable conformations pR and pS could interconvert by oxygen-through-the-annulus rotations of their hydroquinone rings (Fig. 1a)17,18,19,20. Hence, pillar[n]arenes could be considered as promising rotor platforms.

Fig. 1: Overview of controllable rotor platforms based on WP5 skeletons.
figure 1

a Illustration of planar chirality transformation of WP5-M. b 1H NMR spectra of WP5-M in D2O (10.0 mM, 298 K), peaks of methylene groups are shown in green. c Cartoon Illustrations of cation-modulated switchable rotary motion of WP5-M (silver and purple balls represent different counter cations with distinct size). For simplification, only one rotary phenolic ring of WP5 is shown.

Recently, a variety of strategies have been established to lock the planar chirality of pillar[n]arenes, such as guest21,22,23,24, solvent25,26,27, temperature28,29,30, as well as redox31. However, these strategies were based on thermodynamic control, while the research on kinetic control of transformation was rarely reported. Steric hindrance and intramolecular hydrogen bonds have been found crucial impacts on rotational barriers of hydroquinone units in pillar[5]arenes, which were reported by Ogoshi and Stoddart, respectively32,33. Nevertheless, the kinetic control on rotations in pillar[n]arenes, and in particular, the behavior of switchable rotary motions of them was unexplored so far.

Sodium carboxylate pillar[5]arene (WP5-Na) and ammonium carboxylate pillar[5]arene (WP5-NH4) (Fig. 1a), are widely used in aqueous supramolecular assembly systems34,35,36,37. Hitherto, the effect of counter cations on WP5 had never been reported. According to the reported 1H NMR spectra of WP5-Na and WP5-NH4 in D2O34,35,36,37, we serendipitously noticed that the signal of methylene group in the rims of WP5-Na was split into two doublets, while in WP5-NH4 case it was a singlet (marked in green, Fig. 1b), indicating slower rotations of hydroquinone moieties in WP5-Na and faster rotations in WP5-NH438. We supposed such distinct kinetic properties may attribute to impacts of counterions Na+ and NH4+ on rotational barriers of hydroquinone rings in WP5. This discovery provided us possibilities to construct switchable rotations in WP5 by means of cation switch. In addition, it is worth noting that one of the working mechanisms of the aggregation-induced emission (AIE) phenomena was the restriction of intramolecular rotations (RIR, e.g., tetraphenylethylene derivatives), of which the aggregation could reduce the speed of intramolecular rotations, lowering energy dissipations, and thus the fluorescence intensity could be enhanced11. Moreover, pillar[n]arenes have been recently reported as AIEgens (luminogens exhibiting AIE attributes), and their AIE properties were arisen from restricted rotations of their phenolic rings39,40. Inspired by that, we inferred controlling rotary speed of hydroquinone rings in WP5 would thus result in fluorescence switch.

In this work, we investigate the effect of various counterions on rotations of hydroquinone rings in WP5, establishing a controllable rotary system, and its applications in fluorescence switch and anti-counterfeiting inks are further developed (Fig. 1c and Supplementary Movie 1).


Design and synthesis

To explore the impact of counterions on rotations of hydroquinone units in anionic pillar[5]arenes, carboxylate pillar[5]arenes with various monovalent counterions were chosen (WP5-M, Fig. 1a), which could be completely ionized in water. In this work, we focused on alkali metal ions (i.e., Li+, Na+, K+, Rb+, Cs+), and ammonium ion (NH4+) was also included (Fig. 1a). These anionic pillar[5]arenes were facilely synthesized by reacting the carboxylic-substituted pillar[5]arene (P5-COOH) with the hydroxide of corresponding cations (Fig. 2 and Supplementary Figs. 211). Comparing the 1H NMR spectra of WP5-M in D2O at 298 K, it was found that methylene peaks in the rims of WP5-Li, WP5-Na, WP5-K, WP5-Rb, and WP5-Cs were split into two doublets, indicating their non-equivalency (i.e., slow rotations), while WP5-NH4 exhibited a singlet peak, referring to the fastest rotation among WP5 derivativities (Fig. 1b).

Fig. 2: General synthetic routes of WP5-M.
figure 2

WP5-M was obtained by reacting the hydroxide of corresponding cations (MOH) with P5-COOH, the hydrolysis product of P5-COOEt.

Variable-Temperature NMR studies

For quantitatively examining rotational barriers of rotors, WP5-M were then subjected to variable temperature (VT) NMR studies32,33,38,41,42. The two doublet methylene peaks tended to coalesce at 343 K, 338 K, 333 K, 333 K, and 328 K for WP5-Li, WP5-Na, WP5-K, WP5-Rb, WP5-Cs in D2O, respectively (Fig. 3). A rotational barrier (ΔG) of 18.23, 17.55, 15.95, 15.68, and 15.06 kcal/mol was revealed based on Eyring plots38, of which the accuracy was validated by three independent measurements (Table 1 and Supplementary Figs. 1221). This result suggested that counter cations of WP5-M affected rotational barriers of hydroquinone rings in WP5-M indeed. Interestingly, a linear line was obtained when we tried to correlate the experimental rotational barriers with the radius r of corresponding cations (R2 = 0.99, Fig. 4a). It decreases in rotational barrier ΔG as the ionic radius increases (negative slope). Li+, with a smallest radius of 0.76 Å, gave the largest ΔG (i.e., slowest rotation speed). Although it was a failure to measure the rotational barrier of WP5-NH4 owing to the limitation of the freezing point of deuteroxide, a qualitative result could be drawn that WP5-NH4 possessed the lowest rotational barrier.

Fig. 3: Partial VT NMR spectra of WP5-M in D2O.
figure 3

The coalesce temperature for WP5-Li (a), WP5-Na (b), WP5-K (c), WP5-Rb (d), and WP5-Cs (e) was 343 K, 338 K, 333 K, 333 K, and 328 K, respectively (marked with purple dashed). The concentration of rotors was 10.0 mM.

Table 1 The summery of kinetic parameters of WP5-M in D2O
Fig. 4: Kinetic analysis of rotations in WP5-M.
figure 4

a Correlation of the rotational barrier ΔGexp with corresponding cation radius. b Enthalpy-entropy compensation plot of WP5-M. c Illustration of electron repulsion between carboxylate anion (represented with pink balls) and electron-rich cavity (represented with red translucent hemisphere) versus cation (represented with silver balls) lubricant effect.

We inferred that the decreasing tendency of rotational barriers of WP5-M could be ascribed to energy difference of ground states (GS) or transition states (TS), which was induced by counter cations of WP5-M (Supplementary Fig. 22). The impact of GS energy on rotational barriers should be minor, since NH4+ possessed stronger stability to WP5 anion than K+ (contribution of hydrogen bonding) in ground states, but the rotational barrier of WP5-NH4 was much lower than that of WP5-K (Supplementary Fig. 23). Besides, a linear relationship between experimental ΔS and ΔH was observed, which was a typical enthalpy-entropy compensation (R2 = 1.00, Fig. 4b)43. It suggested the existence of enhanced binding force (enthalpy favored) in TS from WP5-Li to WP5-Cs, which would lead to reduction of configurational freedom (entropy decrease). Therefore, we speculated that this downward trend of rotational barriers could be interpreted as following: in rotary process, there was a strong electronic repulsion between carboxylate anions in the rims and the electron-rich cavity of pillar[5]arenes (Fig. 4c). Cation acting as grease, could insert between anions and cavities, involving coulomb interactions with anions and cation-π interactions with cavities, by which the disfavored repulsion could be impaired. Ascribing to a size-specific mechanism that was ubiquitous in macrocyclic host-guest systems (e.g., the binding constant of 18-crown-6 with K+ was higher than that with Na+ and the size of K+ that was more matched with the cavity of 18-crown-6)44,45,46, larger counter cations might have stronger interactions, leading to an enhanced lubricant effect (Fig. 4c). Furthermore, although radii of NH4+ and K+ were similar (1.43 Å and 1.38 Å, respectively), WP5-NH4 had a lowest rotational barrier, which could be attributed to additional hydrogen bond between ammonium ion and carboxylate oxygen. To preliminarily validate this size-specific mechanism in pillar[5]arenes, we measured binding constants (Ka) of M+ with ethoxycarbonylmethoxy-substituted pillar[5]arene (P5-COOEt, Fig. 2) in DMF/H2O solution (v/v = 4/1) using titration method (Supplementary Fig. 26)47,48. Although it was failed to measure Ka of Li+ and Na+ due to the fact that the changes of their UV absorption were too small to conduct the non-linear curve-fitting, an increased Ka of K+, Rb+, and NH4+ was successfully estimated to be 4.2 (± 1.3) × 102, 7.9 (± 1.2) × 102, and 1.1 (± 0.1) × 103 M−1 respectively, indicating an enhanced binding force between cations and cavities of pillararenes, which was consistent with our hypothesis.

P5-COOH, the precursor of WP5-M, was almost insoluble in aqueous solution, and the ionization of its carboxyl hydrogen is incomplete. Notably, Ogoshi reported that the rotational barrier of P5-COOH in DMF-d6 was 17.08 kcal/mol−1 49, which was lower than WP5-Li and WP5-Na but higher than WP5-K, WP5-Rb and WP5-Cs. This may be ascribed to the steric favor to hydrogen than Li+ and Na+, although Li+ and Na+ had modest electronic favor in rotation process. For K+, Rb+ and Cs+, the stabilizing effect brought by electronic effect would overcome the unfavorable steric hindrance. VT-NMR study on P5-COOH in a mixed solvent of DMF-d6/H2O (v/v = 3/1, Supplementary Figs. 2425) was performed. The addition of water in organic solvent would facilitate ionization of carboxyl group (e.g., pKa value of acetic acid in water and DMSO were 4.76 and 12.60 respectively)50, so the electronic effect during rotations of P5-COOH in the above mixed solvent could be amplified. A rotational barrier of 18.37 kcal/mol−1 was revealed for P5-COOH, which was the highest energy barrier among WP5 derivatives. This result suggested that if the carboxyl group could be fully ionized, the lubricant effect of the proton could be very weak owing to its smallest size compared with metal cations.

Theoretical computations

Theoretical calculations were next carried out for validating our above-mentioned hypothesis. To begin with, the binding energy between cations and GS structures of WP5 anion (pS conformation) was calculated (EGS, Supplementary Table 1). EGS fell off in the sequence of NH4+ > Li+ > Na+ > K+ (i.e., the stability sequence of GS: WP5-NH4 > WP5-Li > WP5-Na > WP5-K), which was inconsistent to experimental results. Hence, the difference among GS energy of WP5-M had minor effect on their discrepant rotational barriers. Then, we focused on deciphering transition states of WP5-M during rotations. Considering that the actual transition states of WP5-M may be highly distorted and hardly calculated51, Stoddart’s TS model was employed, where the transformation between pS and pR conformers of WP5-M could be dissected into four plausible pathways with serval possible intermediates (Fig. 5a and Supplementary Fig. 27)33. The potential energy surface scanning (PESS) was subsequently performed to monitor energy variation during rotations (Supplementary Fig. 28)52,53. The stage TS1 was examined primarily (Fig. 5a), in which only one phenolic ring (marked in orange) was allowed free rotation around the cross-section of WP5 (plane α, Fig.5a). Due to the complexity of structures of WP5 derivatives, it was difficult to locate transition state structures in rotation process. Instead, configurations with the highest energy sampled from PESS were considered as preliminary transition state structures for estimating rotational barriers (Supplementary Fig. 29)54,55,56. The calculation results were presented in Supplementary Figs. 3033 and Supplementary Table 2, from which WP5-Li was found to have the highest rotational barrier of 8.83 kcal/mol for TS1. Additionally, the binding energy between cations and TS1 structures of WP5 anion was also computed (ETS, Supplementary Table 1). For the same cation, ETS was higher than its EGS, which showcased stronger impact of cations on transition states than that on ground states.

Fig. 5: Mechanism studies of rotation process of WP5-M.
figure 5

a Partial pathways of transformation between its pS and pR conformers of WP5-M. b Configurations extracted from potential energy surface scanning when hydroquinone ring rotated to the plane α. The dashline represents the distance between cations and corresponding functional groups. The non-bonded distances are in units of Å.

To further understand rotation processes of WP5-M, the energy barrier for TS2 was also evaluated. In this step, hydroquinone rings in ortho- (o-, TS2a) or meta- (m-, TS2b) position of initially rotated unit flipped around plane α (marked in blue, Fig. 5a). It was found that rotational barriers of TS2a were higher than TS2b in all studied systems (e.g., TS2a: 16.18 kcal/mol; TS2b: 13.84 kcal/mol for WP5-Li, Supplementary Figs. 3033, Supplementary Table 2), which may be ascribed to steric hinderance of two neighboring hydroquinone rings in TS2a. Thus, TS2b was a favored pathway when the second hydroquinone ring tended to rotate. The highest energy barriers among two rotational processes (TS1 and TS2b) were evidently reduced from WP5-Li, WP5-Na, WP5-K, to WP5-NH4, which was in accordance with experimental results. Sampled structures from PESS were shown in Fig. 5b, in which hydroquinone rings rotated to the plane α for WP5-Li, WP5-Na, WP5-K, and WP5-NH4, respectively. A decrease in the non-bonded distance between cations and vertical phenolic rings in m- position (WP5-Li: 2.87 Å; WP5-Na: 2.74 Å; WP5-K: 2.35 Å; and WP5-NH4: 2.37 Å) indicated the enhanced cation-π interactions (Fig. 5b). Meanwhile, hydrogen bonding between ammonium ion and carboxylate oxygen of WP5-NH4 was significant, which could be contributed to the lowest rotational barrier of WP5-NH4 among these anionic pillar[5]arenes (Fig. 5b).

Solvent effects

Considering cations are highly solvated in water, and solvation might get involved in the rotation process of WP5-M, WP5-Na was chosen as the model to explore the effect of ionic solvation on rotations. We performed additional VT NMR studies in mixed solvent (D2O mixed with polar protic solvent methanol-d4, or polar aprotic solvent DMSO-d6), due to insolubility of WP5-M in nonaqueous solvent. Rotational barriers and relevant kinetic parameters of WP5-Na in the mixed solvent were equivalent to that in deuterium oxide (Supplementary Table 3 and Supplementary Figs. 3443). This result indicated that the ionic solvation had little effect on rotation process, which may be ascribed to narrow cavities of WP5 limited the solvation of cations in transition states. Since solvents were irrelevant to the rotational barriers of WP5-M, we attempted to study kinetic traits of WP5-NH4 in a mixed solvent (D2O/ methanol-d4, v/v = 3/2, the maximum mixing ratio that be applied for dissolving WP5-NH4 with a 10.0 mM concentration), of which the freezing point was about 228 K. Although a coalesce temperature of 268 K was revealed for WP5-NH4, unfortunately, the lower temperature couldn’t lead the coalescent methylene peaks to fully split into two doublet (Supplementary Fig. 44). Hence, the rotational barrier of WP5-NH4 could not be calculated according to current VT NMR data.

Cation grease/brake rotations and fluorescence switch

Having extensively explored the impact of cations on rotational barriers of WP5, attention was subsequently turned to construct switchable rotary motions in WP5 platforms, where counter cations could grease/brake of rotors. NH4+ and Li+ were chosen for an exchange pair, between which the switch could be easily achieved (Fig. 6a). Two doublets methylene peaks of WP5-Li were observed in 1H NMR spectra (Fig. 6b), showing slow rotations of rotors. NH4F was added for depositing Li+ with F-, leading to an acceleration of rotors (Fig. 6b). Subsequently, the solution was treated with LiOH, heating and bubbling with argon to remove NH3, resulting in a recovery of slow rotations (Fig. 6b). More importantly, the switch of the rotational speed could recur with multiple cycles (Fig. 6c). Furthermore, an attempt of reversibly switching WP5-NH4 and WP5-Na was also achieved, in which Na+ could be captured with 15-crown-5, and NH4+ could be removed under heating and bubbling the solution with argon after basified by NaOH (Supplementary Figs. 4546).

Fig. 6: Rotational speed control and fluorescence switch of WP5-M.
figure 6

a Cation switch between Li+ and NH4+. b Partial NMR of switching control of rotors. c Switching of rate for WP5 in D2O at 298 K with multiple cycles of cation exchange. d Fluorescence emission spectrum of WP5-M (c = 5.0 mM, λex = 365 nm; insert: fluorescence of WP5-Li and WP5-NH4 under 365 nm UV irradiation). e Fluorescence response of WP5 (5.0 mM) at 471 nm upon multiple cycles of cation switch between Li+ and NH4+ in D2O (λex = 365 nm). f Revealing and erasing information with WP5 inks: the image written with WP5-NH4 under 365 nm light (I), and the image could be observed upon adding LiOH on the pre-existing images (II), which could be erased by painting NH4F (III).

Encouraged by these findings, next we put our effort on modulating photochemical properties of WP5. Strikingly, a downward trend of fluorescence intensities of WP5-M under 365 nm irradiation was observed (Fig. 6d), which fall in line with the sequence of rotational barrier. WP5-Li, with the highest rotational barrier, had the strongest luminescence, while s quite weak fluorescence was observed for WP5-NH4 (Fig. 6d). Presumably, restricted rotations reduced energy dissipations, and thus enhanced emission intensity, which was similar to RIR mechanisms of AIE phenomenon11. Furthermore, reversible fluorescence switching with several cycles was accomplished via the exchange of Li+ and NH4+ (Fig. 6e). This controllable fluorescence switch was promising to be considered in anti-counterfeiting technology (Fig. 6f). WP5-NH4 aqueous solution was utilized as inks, and a number “7” could be written on filter paper, which was invisible under 365 nm UV irradiation (Fig. 6f (I)). When applying LiOH aqueous solution on to the surface of pre-writing “7” image, the number could be revealed under 365 nm irradiation (Fig. 6f (II)). Moreover, the visible “7” image could be erased by painting NH4F solution onto the image surface again (Fig. 6f (III)).


In summary, the impact of cations on rotational barriers of anionic pillar[5]arenes has been investigated via dynamic NMR technology. It is found that cations had significant effects on rotational barriers of hydroquinone units in WP5, which is correlated with their radius. Next, theoretical computation is employed for providing mechanistic insights into rotation process. Due to orderly enhanced cation-π interactions between cations and electron-rich cavities in rotary processes, rotational barriers of WP5-M decrease from Li+ to Cs+. WP5-NH4 has the lowest rotational barrier among WP5-M, which is ascribed to additional hydrogen bonding. Finally, switchable rotary motions of rotors are explored. The accelera-tion/deacceleration of rotors is accomplished by means of cation exchange, and these findings further allow the fluorescence switch, which is applied as anti-counterfeiting inks. The strategies and results presented here should find potential applications in many fields, such as sensing, molecular devices, and smart materials.


Dynamic NMR studies

Rotational barriers of rotors WP5-M were measured via variable temperature (VT) 1H NMR by monitoring the broadening of diastereotopic methylene protons in the rims of WP5-M. VT 1H NMR spectra were recorded on a BRUKER AVANCE III 400 MHz or BRUKER AVANCE III 600 MHz spectrometer. The concentration of rotors were 10.0 mM. All VT experiments were performed three times. The NMR line shape analyses were performed on MestReNova 14 sofeware package (Mestrelab Research S. L.). Rates of exchange kex were obtained by line width analysis with the equation:

$${k}_{{ex}}={{{{{\rm{\pi }}}}}}({{{{{\rm{h}}}}}}-{{{{{{\rm{h}}}}}}}_{0})$$

where h represented the width of peak at half height, h0 represented the width of peak at slow or no exchange38.

Kinetic parameters were obtained by using the exchange rates (kex, s−1) obtained from line width analysis of the VT 1H NMR spectra, the enthalpy change (ΔH) and entropy change (ΔS) of the transition state were calculated from Eyring plots:

$${{{{\mathrm{ln}}}}}\frac{k}{{{\mbox{T}}}}=-\frac{\varDelta {H}^{{{\ddagger}} }}{R}\frac{1}{{{\mbox{T}}}}+\frac{\varDelta {S}^{{{\ddagger}} }}{R}+{{{{\mathrm{ln}}}}}\left(\frac{{k}_{B}}{h}\right)$$

where k is the exchange rate constant, T is the absolute temperature, ΔH is the enthalpy of activation, R is the universal gas constant, kB is the Boltzmann constant, h is the Planck’s constant, and ΔS is the entropy of activation. The free energy of activation (ΔG) was calculated through Gibbs equation:

$$\it {\Delta {{{{G}}}}}^{{{\ddagger}} }={\Delta {{{{H}}}}}^{{{\ddagger}} }-{{{{{{\rm{T}}}}}}\Delta {{{{S}}}}}^{{{\ddagger}} }$$

The relevant NMR spectra and Eyring Plots were avalilble at the Supplementary Information.

Theoretical calculation

All computational calculations were carried out with Gaussian 16 software57. The potential energy surface have been scanned using semi-empirical PM6 method with dispersion correction (PM6-D3)52,53. The sampled configurations from the scanned structures were further optimized using the B3LYP-D3 functional58,59,60,61. The standard 6–31 G(d) and 6–31 G + (d) basis sets were used for nonmetal atoms and metal Li, Na and K, respectively. For detail, please see Supplementary Information and Supplementary Data 1.