A Halogen Bonding [2]Rotaxane Shuttle for Chloride-Selective Optical Sensing

Graphical Abstract

A dynamic halogen bonding multi-station [2]rotaxane displays a selective optical response to chloride over bromide and iodide anions. Chloride binding induces a translocation of the isophthalamide-functionalised macrocycle from the naphthalene diimide (NDI) stations to the halogen bonding station, leading to a reduction in the donor-acceptor charge-transfer absorption arising from macrocycle-NDI interactions, whereas the larger halides are excluded from the interlocked cavity.

Abstract

The first example of a [2]rotaxane shuttle capable of selective optical sensing of chloride anions over other halides is reported. The rotaxane was synthesised via a chloride ion template-directed cyclisation of an isophthalamide macrocycle around a multi-station axle containing peripheral naphthalene diimide (NDI) stations and a halogen bonding (XB) bis(iodotriazole) based station. Proton NMR studies indicate the macrocycle resides preferentially at the NDI stations in the free rotaxane, where it is stabilised by aromatic donor-acceptor charge transfer interactions between the axle NDI and macrocycle hydroquinone moieties. Addition of chloride ions in an aqueous-acetone solvent mixture induces macrocycle translocation to the XB anion binding station to facilitate the formation of convergent XB⋅⋅⋅Cl⁻ and hydrogen bonding HB⋅⋅⋅Cl⁻ interactions, which is accompanied by a reduction of the charge-transfer absorption band. Importantly, little to no optical response was induced by addition of bromide or iodide to the rotaxane, indicative of the size discriminative steric inaccessibility of the interlocked cavity to the larger halides, demonstrating the potential of using the mechanical bond effect as a potent strategy and tool in chloride-selective chemo-sensing applications in aqueous containing solvent environments.

Introduction

The prevalence of anions in biological,¹ environmental² and industrial³ processes has driven rapid advances in the field of anion recognition.⁴ Of particular importance is the expanding repertoire of non-covalent interactions utilised in supramolecular anion binding motifs, including hydrogen bonding,⁵ halogen bonding,⁶ chalcogen bonding⁷ and anion-π interactions,⁸ which facilitates the design of receptors containing multiple recognition motifs to achieve potent and selective binding. Saliently, this lays the groundwork for the development of improved anion sensors capable of providing a real-time response to the presence of specific anions. Of the established sensing methods, optical spectroscopy is especially useful due to its sensitivity, fast read-out and technical simplicity.⁹

Mechanically interlocked molecules (MIMs) such as rotaxanes and catenanes have emerged as promising candidates in the development of supramolecular hosts and, via signalling group integration, as chemo-sensing materials.¹⁰ From a molecular recognition perspective, the encapsulation of a target guest within the three-dimensional interlocked cavities of MIMs, which can be strategically functionalised with recognition motifs, is a well-established strategy to enhance both the affinity and selectivity of guest binding.¹¹ Crucially, in the context of sensor development, the dynamic intercomponent co-conformational movement available to MIMs has the potential to be exploited as a reversible stimuli-responsive novel sensing mechanism. However, despite substantial research efforts directed towards the synthesis of light-,¹² electrochemical-¹³ and pH-switchable¹⁴ rotaxane shuttles, there are surprisingly few reports of MIM-based sensors that undergo co-conformational changes in response to anion binding.¹⁵

The pivotal role played by chloride¹⁶ and iodide^1b in various aspects of human health provides an impetus for the development of chemo-sensors capable of distinguishing these chemically-similar halide anions. However, achieving selective chloride over iodide recognition and sensing in aqueous-containing media remains a persistent challenge¹⁷ due to the extensive solvation of chloride by water molecules, which typically confers a “Hofmeister bias” in halide binding affinities, i. e. Cl⁻<Br⁻<I⁻.¹⁸

Herein we report a multi-station halogen bonding (XB) [2]rotaxane shuttle 1 which functions as a selective optical sensor for chloride over the larger halide anions. The host design involves an isophthalamide-based macrocycle threaded onto a 3-station axle containing naphthalene diimide (NDI) stations as optical reporter groups, as well as a central bis(iodotriazole) based halogen bonding (XB) station for anion binding (Figure 1). Through detailed NMR investigations, we demonstrate the macrocycle resides preferentially over the peripheral NDI stations in the free rotaxane – a co-conformation stabilised by the formation of aromatic donor-acceptor charge transfer interactions between the electron-deficient NDI stations and the electron-rich hydroquinone moieties of the macrocycle. ¹H NMR anion binding studies in an aqueous-acetone solvent mixture indicate chloride-induced macrocycle translocation to the central XB station, which is driven by the formation of convergent XB⋅⋅⋅Cl⁻ and HB⋅⋅⋅Cl⁻ interactions with the axle and macrocycle respectively. UV-vis spectroscopic studies show a reduction in the intensity of the characteristic charge-transfer absorption band in response to chloride binding, corresponding to the loss of mechanical bond intercomponent NDI−hydroquinone interactions. Crucially, this optical response is not elicited by bromide and iodide binding due to the steric inaccessibility of the interlocked cavity to the larger halides, thereby demonstrating for the first time a rotaxane-based optical sensor capable of responding selectively to chloride.

Figure 1

Schematic representation of 3-station-[2]rotaxane 1 acting as a selective optical sensor for chloride over iodide anions.

Results and Discussion

Synthesis and Characterisation

Adapting our recent preparation of XB [2]rotaxanes¹⁹ and hetero[2]catenanes²⁰ via an anion template-directed route, we sought to employ a similar strategy to prepare a multi-station [2]rotaxane. The rotaxane design consists of an axle containing peripheral NDI stations and a central 4,6-dinitro-1,3-bis(iodotriazole) halogen bonding donor motif, as well as an isophthalamide-based macrocycle. To this end, the 3-station axle 2 was synthesised in 90 % yield via a Cu(I)-catalysed azide−alkyne cycloaddition (CuAAC) reaction between the bis(iodoalkyne) 5 and two equivalents of NDI-functionalised stopper azide 4 (Scheme 1). With the axle in hand, the target [2]rotaxane 1 was prepared via a chloride anion-directed amide condensation cyclisation reaction between bis(amine) 6 and 3,5-bis(chlorocarbonyl) pyridine 7 to form the interlocked rotaxane product 1 (Scheme 1). In a typical reaction, a solution of freshly-prepared bis(acid chloride) 7 was added dropwise to a cooled solution of the axle 2, tetrabutylammonium chloride (TBACl), bis(amine) 6 and triethylamine in anhydrous CH₂Cl₂. The reaction mixture was allowed to warm to room temperature and stirred overnight, then subjected to a basic work-up and purification by iterative preparative TLC to afford the desired [2]rotaxane 1 as a bright orange solid in an isolated yield of 11 %, which was characterised by ¹H NMR, ¹³C NMR and high-resolution mass spectrometry.

Scheme 1

(a) Synthesis of axle 2 and 3-station-[2]rotaxane 1 via a chloride anion template directed strategy. (b) Synthesis of halogen bonding axle 3.

A comparison of the ¹H NMR spectrum of 1 to those of the non-interlocked axle 2 and macrocycle 8 components provides evidence for mechanical bond formation (Figure 2). The hydroquinone protons of the macrocycle H_6,7 undergo a dramatic upfield shift (Δδ=1.06 ppm), indicating the formation of aromatic stacking interactions with the aryl components of the axle. Pronounced upfield shifts were also observed in the protons of the tetra(ethylene glycol) linker H_8-11, presumably due to their increased proximity to the shielding ring currents of the axle aryl groups, while the internal pyridyl proton H₂ shifts downfield in a manner consistent with previous reports of interlocked structures featuring an isophthalamide-based macrocycle.^{15b, 15d, 21} The changes in the chemical shift values of the axle protons are considerably more modest, with the most prominent changes being a downfield shift of the NDI protons H_i,j (Δδ=0.07 ppm). This is in contrast to previously reported structurally-comparable [2]rotaxane shuttles,^{15b, 15d} in which the NDI proton signals were shifted upfield by a considerable margin when sandwiched between two hydroquinone groups. This suggests that other competing factors influence the observed chemical shift values of rotaxane 1, particularly given the conformational flexibility of the axle which may promote the formation of secondary intercomponent contacts.

Figure 2

Overlaid ¹H NMR spectra of HB macrocycle 8 (top), [2]rotaxane 1 (middle) and axle 2 (bottom) (500 MHz, acetone-d₆, 298 K).

Further insights into the co-conformation of 1 were obtained from the 2D ¹H-¹H ROESY spectrum (Figure 3). Strong through space cross-peaks were observed between the NDI protons H_i,j and several macrocycle-derived proton signals in the spectrum, most notably the hydroquinone protons H_6,7, suggesting the formation of π-stacking interactions between the NDI and hydroquinone moieties. Significant intercomponent cross-peaks were also observed between H_6,7 and the NDI-adjacent biphenyl protons H_g,f, while no cross peaks were observed between the dinitrobenzene protons H_a and H_b and the macrocycle. This is consistent with the macrocycle displaying a co-conformational bias towards residing at the peripheral regions of the axle rather than the central halogen bonding station. This co-conformation is expected to be stabilised by the formation of favourable aromatic charge-transfer interactions between the electron-deficient NDI and the electron-rich hydroquinone moieties, as suggested by the appearance of a characteristic charge-transfer absorption at λ=450 nm (Figure 4). The presence of weak cross-peaks between the hydroquinone protons H_6,7 and the proximal biphenyl protons H_g,f further supports the preferential occupation of the axle-based NDI stations by the macrocyclic component.

Figure 3

¹H-¹H 2D ROESY NMR spectrum of 3-station-[2]rotaxane 1 (600 MHz, acetone-d₆, 298 K).

Figure 4

¹H NMR spectroscopic anion binding studies of rotaxane 1 with halide guests. (a) Truncated stacked ¹H NMR spectra of 1 showing progressive additions of TBAX, where X=Cl (left), I (right). [Rotaxane]=0.5 mM. Solvent=2.5 % D₂O/acetone-d₆, T=298 K. (b) Cartoon representations of the proposed endotopic and exotopic binding modes adopted by Cl⁻ (left) and I⁻ (right) respectively.

¹H NMR Anion Binding Studies

¹H NMR spectroscopic binding studies were conducted to investigate the effect of halide binding on the co-conformation of rotaxane 1. In a typical experiment, aliquots of TBAX salts (X=Cl, Br, I) were added progressively to a 0.5 mM solution of the host in the moderately competitive 2.5 % D₂O/acetone solvent mixture. To determine the anion affinities of the XB donor-containing axle motif, solubility constraints of the NDI-containing axle 2 required the preparation of a model axle 3. Structurally similar to 2, axle 3 does not feature the NDI units, which are not anticipated to influence the anion binding properties of the XB donor unit. Significant rotaxane chemical shift perturbations were observed in response to increasing halide concentration, particularly in the 2,4-dinitrobenzene protons H_a and H_b, indicative of anion binding at the central XB motif of the axle component (Figure 3a). Pleasingly, the magnitude and direction of the changes, particularly of the proton signals proximal to the macrocycle HB binding site, differed significantly with the identity of the halide guest, suggesting that the halides adopt different binding modes to rotaxane 1. Importantly, this behaviour was not observed in axle 3, which consistently displayed an upfield movement of H_a and downfield movement of H_b in response to all three halides, providing strong evidence that the variation in anion binding modes seen in 1 arises from a mechanical bond effect.

Addition of chloride anions results in pronounced downfield shifts of the rotaxane's internal pyridyl proton H₂ and amide protons H₃, indicating that the macrocycle component HB binding motif is involved in anion binding (Figure 4a). Taken together with the changes in the proton signals near the XB binding site, namely a downfield movement of axle proton H_a and significant broadening of H_b, this suggests that the small chloride anion is convergently bound by the XB donors of the axle and the isophthalamide group of the macrocycle in 1 (Figure 4b). Similar chemical shift perturbations in response to chloride binding have been observed in a series of [2]rotaxanes and [2]catenanes containing an analogous binding site, which is capable of encapsulating chloride in a binding cavity formed by the interlocked components.^{19, 20} In rotaxane 1, the adoption of this endotopic binding mode necessitates a translocation of the macrocycle to the central XB bis(iodotriazole station), as indicated by a large upfield shift of the iodotriazole-adjacent methylene singlet H_c, which presumably experiences greater shielding by virtue of its increased proximity to the ring currents of the macrocycle hydroquinone groups. Significant chemical shift movements were also observed in the macrocycle protons H_4,5 and H_8-11, which are distal from the anion binding site, thereby suggesting that chloride binding additionally induces a co-conformational change in the host.

In contrast, little to no changes to the macrocycle proton signals were observed upon addition of iodide (Figure 4a). The axle proton H_a moves downfield with increasing concentrations of iodide, displaying a similar response to that of 3 and other non-interlocked anion receptors containing the 2,4-dinitrobenzene-bis(iodotriazole) XB motif.¹⁹ This suggests that iodide is bound primarily by XB⋅⋅⋅anion interactions with the iodotriazole XB donors of the axle, with minimal participation of the isophthalamide group of the macrocycle (Figure 4b). This is consistent with larger ionic radius of the iodide anion precluding access to the interlocked binding cavity of the rotaxane and forcing it to adopt an exotopic binding mode. Importantly, no significant movements were observed in proton signals distal from the XB binding site, suggesting that the co-conformation of 1 does not change significantly in response to iodide binding.

Addition of bromide produces a chemical shift perturbation profile which appears to be intermediate between the extreme cases of chloride and iodide (Figure S15). With increasing bromide concentration, the macrocycle protons H₂ and H₃ move progressively downfield (similar to the chloride binding experiments), while axle proton H_a moves upfield (similar to the iodide binding experiments), albeit with greatly attenuated magnitudes relative to the corresponding chloride/iodide-induced changes. Given that the ionic radius of the bromide anion falls in between that of the other two halides, it is conceivable that the 1⋅Br⁻ complex in solution exists as a mixture of the competing endotopic and exotopic binding modes in fast exchange.

The anion association constants of rotaxane 1 and axle 3 in 2.5 % D₂O/acetone-d₆ were subsequently determined by a global Bindfit²² analysis of the titration data (Figure S20–21), which determined 1 : 1 stoichiometric host-guest association constants for chloride and iodide binding, summarised in Table 1. In contrast, the complex scenario indicated by the bromide-induced chemical shifts of 1, with qualitative ¹H NMR analysis suggesting both endotopic and exotopic binding of bromide, is challenging to model from the titration data and prohibits a reliable determination of bromide association constants. Such a scenario highlights the need for judicious consideration of all possible equilibria as well as the underlying chemical structures when attempting to extract quantitative information from a host-guest binding experiment, which is expected to become more important with the development of increasingly large and complex supramolecular systems.²³

Table 1. Halide anion association constants (K_a/M⁻¹) for rotaxane 1 and axle 3 determined by ¹H NMR titrations.^[a]

	Anion Association Constant Ka (M^−1)
Anion	Rotaxane 1	Axle 3
Cl^−	3250(102)	770(12)
Br^−	N.D.^[b]	2370(52)
I^−	2720(201)	2584(86)

• ^[a]K_a values calculated using Bindfit with a 1 : 1 host-guest binding model. Errors (±) are all <5%. All anions added as TBA⁺ salts. Solvent=2.5 % D₂O/acetone-d₆. T=298 K. [Receptor]=0.5 mM. ^[b]Not determined (see text for discussion).

Inspection of the anion association constants summarised in Table 1 reveal trends consistent with the proposed binding modes for the various halides. Considering first the simple 1-station XB axle 3, we observe an increase in anion binding affinities in the order Cl⁻≪Br⁻<I⁻. This is expected on the basis of the Hofmeister series, wherein the most extensively hydrated chloride anion is least strongly bound to the receptor in aqueous containing solvent mixtures.

Notably, this is not the case for rotaxane 1, which displays a significant 4.3-fold enhancement in chloride binding affinity relative to axle 3. Indeed, the chloride association constant of 1 is higher than the corresponding value for iodide, displaying an ‘anti-Hofmeister’ binding preference. This is consistent with the proposed endotopic chloride binding mode in which binding affinity is enhanced by the formation of convergent interactions involving both the XB and HB binding motifs in 1. In contrast, the iodide association constants of rotaxane 1 and axle 3 are similar within experimental error, providing strong evidence that the iodide ion interacts almost exclusively with the XB donor motif of the axle and does not benefit from an enhancement in binding affinity originating from its encapsulation in the interlocked cavity. Crucially, the enhancement in chloride but attenuation of iodide binding affinity of the rotaxane relative to the non-interlocked axle demonstrates the efficacy of using mechanically bonded hosts to achieve highly selective enhancements in guest binding affinities.

Optical Studies

The UV-visible spectrum of the free rotaxane 1 in 2.5 % H₂O/acetone contained a broad absorption at λ=450 nm, diagnostic of donor-acceptor charge-transfer interactions formed between the NDI stations of the axle and the hydroquinone groups of the macrocycle (Figure 5). In contrast, no significant absorption intensity was observed in the non-interlocked NDI-containing axle 2 (Figure S22).

Figure 5

UV-Vis spectra showing the effects of adding 20 equivalents of TBAX (X=Cl, Br, I) into a solution of rotaxane 1 in 2.5 % H₂O/acetone. [Host]=0.25 mM. T=298 K.

Encouraged by the ¹H NMR anion binding studies which indicate a chloride-induced translocation of the macrocycle to the XB binding site, attention was directed towards investigating the optical response of rotaxane 1 towards anion binding. To this end, preliminary UV-vis spectroscopic sensing studies were conducted on a 0.25 mM solution of 1 in 2.5 % H₂O/acetone, wherein the UV-vis absorption spectrum of the host solution was recorded before and after the addition of 20 equivalents TBAX (X=Cl, Br, I) (Figure 5).

Pleasingly, upon addition of chloride anions to rotaxane 1, a significant (42 %) reduction of the absorption intensity at 450 nm was observed, which was accompanied by a subtle discolouration of the bright orange solution, consistent with a reduction in the NDI−hydroquinone charge-transfer interaction resulting from translocation of the macrocycle to the central XB anion binding station. Crucially, the bromide-induced reduction in absorption intensity at 450 nm was considerably smaller (10 %), while iodide addition results in essentially no change in the charge-transfer band. This serves as an effective demonstration of exploiting the binding cavity of an interlocked molecule to simultaneously achieve an enhancement in binding affinity as well as a selective optical sensing response for a target guest ion.

Conclusions

A 3-station XB [2]rotaxane 1 was synthesised by using a discrete chloride ion template to direct the cyclisation of an isophthalamide based macrocycle around a halogen bonding axle. 1D and 2D ¹H NMR analysis of the rotaxane suggested a co-conformation in which the macrocycle resides preferentially at the axle NDI stations, which is driven by the formation of mechanically bonded intercomponent NDI−hydroquinone aromatic donor-acceptor charge transfer interactions. The anion binding properties of rotaxane 1 and a model axle 3 were studied by ¹H NMR titration studies in an aqueous-acetone solvent mixture, revealing a significant enhancement in chloride binding affinity in the rotaxane relative to the axle, which was attributed to the encapsulation of chloride in the interlocked binding cavity, where it interacts convergently with the halogen bonding and hydrogen bonding motifs of the axle and macrocycle respectively. In contrast, no enhancement of iodide binding affinity was observed, presumably due to the inability of the larger iodide ion to fit within the interlocked cavity. A complex binding scenario was observed for bromide, which is rationalised by competition between the endotopic and exotopic binding modes. Importantly, rotaxane translocation of the macrocycle from the NDI station to the XB station induced by chloride binding led to an optical chemo-sensing response corresponding to a significant reduction in the intensity of the charge transfer absorption band. This work illustrates the potential efficacy of using mechanically interlocked host molecules as a strategy to not only enhance the binding affinity and selectivity of anion receptors using the mechanical bond effect, but also through their inherent dynamism achieve selective optical sensing of a target ion over other chemically similar guest species.

Supporting Information

All synthetic procedures, characterization of novel compounds and ¹H NMR titration data are available within the Supporting Information.^{15b, 24}

Acknowledgments

H.M.T thanks the Clarendon Fund and the Oxford Australia Scholarships Fund for a scholarship. A.D. and A.T. thank the EPSRC for studentship funding (Grant reference numbers: EP/N509711/1 and EP/T517811/1).

Conflict of interests

The authors declare no conflict of interest.

Open Research

Data Availability Statement

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