Alkali Metal Halide Ion-Pair Binding in Conformationally Dynamic Halogen Bonding Heteroditopic [2]Rotaxanes

Graphical Abstract

A series of heteroditopic halogen bonding [2]rotaxanes display metal cation binding-induced co-conformational dynamism and significant positive cooperativity upon binding of alkali metal halide ion-pairs. Detailed analysis of the ¹H NMR titration spectra reveal a complex suite of parallel and competing equilibria which need to be considered to obtain reliable estimates of the rotaxanes’ ion-pair binding affinities.

Abstract

A series of heteroditopic halogen bonding (XB) [2]rotaxanes were prepared via a combination of passive and active metal template-directed strategies. The ability of the [2]rotaxanes to bind alkali metal halide ion-pairs was investigated by extensive ¹H NMR titration studies, wherein detailed analysis of cation, anion and ion-pair affinity measurements indicate dramatic positive cooperative enhancements in halide anion association upon either Na⁺ or K⁺ pre-complexation. This study demonstrates that careful consideration of multiple, parallel and competing binding equilibria is essential when interpreting observed ¹H NMR spectral changes in ion-pair receptor systems, especially those which exhibit dynamic behaviour. Importantly, in comparison to XB [2]catenane analogues, these neutral XB heteroditopic [2]rotaxane host systems demonstrated that despite their relatively weaker cation and anion binding affinities, they exhibit a notably higher level of positive cooperativity for alkali metal halide ion-pair binding, highlighting the role of greater co-conformational adaptive behaviour in mechanically-bonded hosts for the purposes of charged species recognition.

Introduction

The design of heteroditopic ion-pair receptors possessing recognition sites for both cation and anion guests is a field of research that has garnered considerable attention in recent years.^1-5 By exploiting the positive cooperativity associated with binding oppositely charged species, ion-pair receptors typically exhibit superior binding properties over monotopic cation/anion receptors, and have consequently demonstrated their efficacy in a multitude of applications including salt extraction/solubilisation,^6-19 membrane transport^20-24 and recognition of biologically-relevant zwitterions.^25-28

The multi-component and topologically complex nature of mechanically interlocked molecules makes them promising scaffolds upon which multiple cation and anion binding sites can be incorporated, facilitating a modular approach to constructing complex heteroditopic and multitopic receptors. However, the design of mechanically interlocked heteroditopic receptors remains an under-explored area, with the handful of reported examples typically utilising hydrogen bonding (HB) as the primary means by which anion binding is achieved.^28-33

We recently prepared a series of halogen bonding (XB) heteroditopic [2]catenanes for the recognition of alkali metal halide ion-pairs.³⁴ The [2]catenanes were prepared via a method pioneered by Chiu et al.,^35-41 exploiting a sodium metal cation to direct the assembly of a pseudo [2]rotaxane complex comprised of an oligo(ethylene glycol)-functionalised XB macrocycle and a bis(azide) precursor. Cyclisation of the latter was achieved via a Cu(I)-catalysed azide-alkyne cycloaddition (CuAAC) reaction to generate the [2]catenane. Motivated by the efficacy of this strategy, we sought to investigate their XB heteroditopic [2]rotaxane analogues by performing a stoppering-type reaction of the intermediate pseudo [2]rotaxane assembly with sterically encumbering terphenyl groups. Indeed, it is noteworthy that in the context of ion-pair molecular recognition by interlocked structures, to the best of our knowledge comparative studies of ‘mechanical bond isomers’ is unprecedented, which is surprising considering the markedly different conformational behaviours of rotaxanes and catenanes and the potential advantages these might endow on ion-recognition properties.

Herein we report a series of halogen bonding heteroditopic [2]rotaxanes containing poly(ethylene glycol) cation binding and XB anion binding sites. (Figure 1). Alkali metal cation ¹H NMR binding studies indicate that the [2]rotaxanes undergo substantial co-conformational changes upon cation recognition, in which the macrocycle shuttles from the axle termini towards the central ethylene glycol station to enable convergent alkali metal cation complexation by the polyether regions of the two interlocked components. Notably, extensive quantitative cation, anion and ion-pair ¹H NMR titration experiments indicate augmented ion-pair recognition properties of the [2]rotaxanes relative to their [2]catenane analogues (Figure 1).

Figure 1

Cartoon representation of heteroditopic [2]rotaxanes studied in this work, highlighting the strong positive cooperativity in ion-pair binding to the metal-complexed rotaxanes. Purple spheres represent halogen bond donor atoms.

Results and Discussion

Synthesis and characterisation of [2]rotaxanes

The all-halogen bonding [2]rotaxanes were synthesised via an alkali metal template-directed double stoppering-based strategy. In a typical reaction, the macrocycles 2⋅XB^DEG or 2⋅XB^TEG were stirred with 1 equiv. M^IBAr₄^F (M = Na⁺ or K⁺) and the corresponding bis(azide) 1^DEG or 1^TEG in CH₂Cl₂ to form pseudo-[2]rotaxane complexes, to which was added 2.5 equiv. stopper iodoalkyne 4, Cu(CH₃CN)₄PF₆ and tris(benzyltriazolylmethyl) amine (TBTA) to facilitate rotaxane formation via a CuAAC reaction. After stirring at room temperature for 24 h, mass spectrometric analysis of the reaction mixtures revealed peaks with m/z = 2565 and 2652, corresponding to the [M+H]⁺ adducts of the target rotaxanes. Following aqueous work-up with EDTA/NH₄OH and purification by preparative thin-layer chromatography, the target rotaxanes 3⋅XB-XB^DEG and 3⋅XB-XB^TEG were obtained in 21 % and 20 % yields respectively (Scheme 1, Route A).⁴²

Scheme 1

Synthesis of [2]rotaxanes via metal cation template-directed pseudo-[2]rotaxane formation (Route A) or Cu(I)-mediated active metal templation (Route B). The chemical structures of the previously-reported XB [2]catenanes³⁴ are included in an inset at the top right.

Interestingly, in stark contrast, employing this method with a proto-alkyne stopper derivative 5 to prepare the mixed HB and XB rotaxanes 3⋅XB-HB^DEG and 3⋅XB-HB^TEG, containing iodotriazole groups in the macrocycle and prototriazole groups in the axle respectively, proved to be unsuccessful. It was observed that when the stopper proto-alkyne was added to the reaction mixture, immediate precipitation occurred, which persisted despite prolonged sonication. After two days, TLC analysis of the reaction mixture showed nearly complete consumption of the bis(azide), but the target [2]rotaxane was not observed on ESI mass spectrometry and purification of the crude yielded only non-interlocked axle product. This may suggest the formation of a highly insoluble Cu(I) acetylide complex which is presumably active towards CuAAC reaction, but incapable of effectively participating in a CuAAC reaction with the Na⁺ templated pseudo-[2]rotaxane complex.

We therefore sought to exploit an alternative method of constructing the target [2]rotaxanes. Motivated by our previous demonstration that iodotriazoles are bifunctional motifs capable of acting as competent ligands for Cu(I) coordination in an active metal template methodology,⁴³ we pre-complexed the macrocycles 2⋅XB^DEG or 2⋅XB^TEG with [Cu(MeCN)₄]PF₆ by stirring at room temperature for 30 min, followed by addition of 4 equiv. bis(azide) 1^DEG or 1^TEG and 8 equiv. stopper alkyne 5. After several days, aqueous work-up and purification by preparative TLC afforded 3⋅XB-HB^DEG and 3⋅XB-HB^TEG in 21 % and 17 % yields respectively (Scheme 1, Route B). It is noteworthy that despite the rotaxane yields obtained being comparable to the alkali metal cation template-directed strategy, in our hands the CuAAC-AMT protocol required a considerable excess of both the bis(azide) and stopper alkyne components as well as longer reaction times and elevated temperatures to achieve commensurate yields (3 days for 3⋅XB-HB^DEG at 80 °C or 7 days for 3⋅XB-HB^TEG at room temperature).

¹H NMR analysis of rotaxane co-conformation

To determine any potential co-conformational changes of the [2]rotaxanes by cation, anion or ion-pair recognition, we initially sought to characterise the interlocked structure's preferred co-conformation in the absence of ion binding events. As the macrocycle and axle components are derived from the same bis(azide) precursor, the ¹H NMR resonances corresponding to H₆-H₁₁₍₁₂₎ (from the axle) and H_d-H_i(j) (from the macrocycle) are challenging to distinguish solely on the basis of chemical shift and multiplicity. As such, 2D ¹H-¹H COSY and ROESY NMR techniques (Figures S3–4, S9–10, S15–16, S21–22) were required to fully assign the ¹H NMR spectra of the four rotaxanes.

Comparing the ¹H NMR spectra of the rotaxanes to their constituent non-interlocked components in 1 : 1 CD₃CN/CDCl₃ (Figures 2, S11, S17, S23) reveals general upfield shifts in the proton resonances originating from both the macrocycle and axle (Table 1), consistent with the shielding effect that typically accompanies mechanical bond formation. In the di(ethylene glycol) (DEG)-based all-XB rotaxane 3⋅XB-XB^DEG (Figure 2), the most prominent upfield shifts in the axle component are observed for xylene protons H₇, which exhibits an upfield shift of 0.54 ppm in the rotaxane 3⋅XB-XB^DEG relative to the free axle 4⋅XB^DEG. Significant upfield shifts were also observed in xylene proton signal H₈ (0.45 ppm) and methylene singlet H₉ (0.35 ppm). In contrast, only modest upfield shifts occurred at the di(ethylene glycol) protons H₁₀ and H₁₁ (0.14 and 0.19 ppm) (Table 1). This suggests that, in the free rotaxane, the macrocycle preferentially resides over the xylene spacers rather than the central di(ethylene glycol) station, which is presumably attributable to favourable π-π interactions between two of the aromatic surfaces of the constituent axle and macrocycle components.

Figure 2

Stacked ¹H NMR spectra of axle 4⋅XB^DEG, rotaxane 3⋅XB-XB^DEG and macrocycle 2⋅XB^DEG (500 MHz, 1 : 1 CDCl₃/CD₃CN, 298 K).

Table 1. Comparison of selected ¹H NMR chemical shifts of rotaxanes and corresponding axles.

Proton	Δδ3/4 (ppm)^[a]
	3/4⋅XB-XB^DEG	3/4⋅XB-HB^DEG	3/4⋅XB-XB^TEG	3/4⋅XB-HB^TEG
H3	0.1598	0.0600	0.0899	0.0522
H4	0.3234	0.1186	0.1843	0.1089
H5	0.1142	0.0773	0.1456	0.1219
H6	0.1986	0.2860	0.1789	0.2125
H7	0.5436	0.7512	0.3678	0.4202
H8	0.4572	0.5098	0.3426	0.4288
H9	0.3466	0.3646	0.2533	0.2989
H10	0.1425	0.0996	0.1215	0.1047
H11	0.1932	0.1842	0.1661	0.1751
H12	–	–	0.0829	0.0637
Htrz	–	0.1365	–	0.1634

• [a] Δδ_3/4 (ppm)=δ(axle) – δ(rotaxane). For example, 3/4⋅XB-XB^DEG refers to δ(4⋅XB^DEG) – δ(3⋅XB-XB^DEG). All ¹H NMR spectra run in 1 : 1 CD₃CN/CDCl₃, 500 MHz, 298 K. Background colours are used to indicate the magnitude of the chemical shift difference. Red: |Δδ|>0.3 ppm; Orange: 0.2<|Δδ|<0.3 ppm; Yellow: 0.1<|Δδ|<0.2 ppm.

In the analogous HB/XB rotaxane 3⋅XB-HB^DEG (Figure S17), the upfield shifts of H₇ (0.75 ppm) and H₈ (0.51 ppm) in the rotaxane relative to the axle 4⋅HB^DEG are even larger, as is the case for H₆ (0.29 ppm), whereas the magnitudes of the upfield shifts in H₁₀ and H₁₁ are slightly diminished (0.10 and 0.18 ppm) (Table 1). This suggests an even stronger preference of the macrocycle for the xylene-triazole region of the axle, which may be due to the propensity for the axle triazole proton to undergo C−H⋅⋅⋅O hydrogen bonding with the di(ethylene glycol) oxygen atoms of the macrocycle.³⁴ This proposed HB interaction between the axle triazole and macrocycle di(ethylene glycol) linker is further supported by the ¹H-¹H 2D ROESY NMR spectrum of 3⋅XB-HB^DEG, which shows significant through-space correlations between H_trz and H_h/i (Figure S16).

In the macrocycle component, the internal benzene proton H_c exhibits a dramatic upfield shift upon rotaxane formation in both 3⋅XB-XB^DEG (1.09 ppm) and 3⋅XB-HB^DEG (1.15 ppm), likely a result of its increased proximity to the shielding ring currents of the aromatic groups in the axle. Interestingly, in the tri(ethylene glycol) (TEG)-based rotaxanes the upfield shifts in H_c, whilst still large, are less pronounced (3⋅XB-XB^TEG: 0.65 ppm; 3⋅XB-HB^TEG: 0.72 ppm), possibly due to the larger macrocyclic cavity which reduces the spectral effects associated with mechanical bond formation.

Guest binding studies

The cation, anion and ion-pair binding properties of the four rotaxanes were studied by ¹H NMR titrations in 1 : 1 CDCl₃/CD₃CN. These conditions are identical to those used previously to investigate the binding properties of the analogous [2]catenanes,³⁴ enabling a direct comparison of ion-pair binding of the rotaxane and catenane topologies.

Upon progressive addition of MBAr₄^F (M = Na⁺, K⁺) to all four rotaxanes, the most prominent changes occurred in the peaks arising from the oligo(ethylene glycol) regions. In particular, the axle protons H₁₀, H₁₁ (and H₁₂ in the TEG-based rotaxanes 3⋅XB-XB^TEG and 3⋅XB-HB^TEG) shifted upfield, indicative of cation binding occurring via the oligo(ethylene glycol) oxygen atoms. In contrast, the corresponding ethylene glycol protons of the macrocycle H_h and H_i (and H_j) exhibited modest downfield shifts. This is tentatively attributed to the competing upfield and downfield shifts induced respectively by cation coordination and macrocycle translocation away from the xylene region of the axle. Significant downfield shifts were also observed in certain axle protons distal from the proposed cation binding site, most prominently xylene protons H₇ and H₈ (in all rotaxanes) and the triazole protons H_trz (present only in 3⋅XB-HB^DEG and 3⋅XB-HB^TEG), while H_4, H₅ and H₆ exhibited smaller downfield movements. Taken together, these spectral changes suggest that the macrocycle translocates to the oligo(ethylene glycol) region of the axle in the presence of a cationic guest, presumably to enable the polyether regions of both interlocked components to convergently bind the cationic guest, highlighting the use of cation binding as a stimulus to induce large amplitude co-conformational changes in [2]rotaxane systems.

The generated binding isotherms from monitoring the chemical shift perturbations of the ethylene glycol region of the axle (H₁₀ and/or H₁₂) were analysed using Bindfit⁴⁴ analysis and fitted to a 1 : 1 host–guest binding model to determine the cation association constants of each rotaxane (Table 2). In general, the TEG-based rotaxanes bind more strongly to both Na⁺ and K⁺ than their respective DEG-based analogues, presumably due to the larger number of polyether oxygen donors available for cation complexation. Similar to the [2]catenane systems, the DEG-based rotaxanes show a preference for the smaller Na⁺ cation while the opposite is true for the TEG-based rotaxanes, reflecting size complementarity between the binding cavity and the preferred cation.

Table 2. Cation binding constants (K_a/M⁻¹) of interlocked receptors determined by ¹H NMR.^[a]

Guest	3⋅XB-XB^DEG	3⋅XB-HB^DEG	3⋅XB-XB^TEG	3⋅XB-HB^TEG
Na^+	216(3) C: 756(11)^[b]	110(1) C: 36(1)^[b]	403(5) C: 753(10)^[b]	244(3)
K^+	69(1) C: 284(2)^[b]	N. B.^[c]	724(11) C: 1734(15)^[b]	329(4)

• [a] K_a values calculated using Bindfit with a 1 : 1 host–guest binding model. Errors are shown in parentheses and are all <2 %. All cations added as BAr₄^F− salts. Solvent=1 : 1 CDCl₃/CD₃CN. T=298 K. [Receptor]=1.0 mM. [b] Where present, the italicised value after “C:” indicates the previously reported K_a of the analogous [2]catenane for the same metal cation. [c] N.B.=No binding observed.

Despite possessing ostensibly identical cation binding sites, the analogous mixed XB/HB rotaxanes display significantly reduced cation affinity relative to their all-XB analogues. The weaker cation binding in the prototriazole-containing rotaxanes is likely due to the energetic cost of disrupting the intercomponent C−H⋅⋅⋅O hydrogen bonds between the axle triazole proton and the macrocycle polyether oxygen atoms (Figure 3) which is consistent with the aforementioned conformational bias of the HB rotaxanes. Similar HB interactions were previously observed in the single crystal X-ray structure of a mixed XB/HB catenane.³⁴

Figure 3

Truncated chemical structure of 3⋅XB-HB^DEG showing postulated intercomponent hydrogen bonding interactions.

Compared to the all-XB catenanes, the all-XB rotaxanes 3⋅XB-XB^DEG and 3⋅XB-XB^TEG display significantly lower alkali metal cation K_a values (Table 2). This suggests a lower degree of pre-organisation in the cation binding sites of the rotaxanes, consistent with the macrocycle having to translocate from the xylene/triazole region of the axle to the central ethylene glycol station to participate in cation binding.

The anion binding properties of the free rotaxanes were investigated via analogous ¹H titration experiments involving the progressive addition of halide anions as their tetrabutylammonium (TBA) salts to solutions of the rotaxanes in 1 : 1 CDCl₃/CD₃CN.

In the all-XB rotaxanes 3⋅XB-XB^DEG and 3⋅XB-XB^TEG, the macrocycle internal benzene proton H_c, which typically exhibits the largest anion binding induced perturbations within the cleft formed by the bis(iodotriazole)benzene XB donor motif, exhibited modest downfield shifts, indicating participation of the macrocycle XB donor groups in anion binding. In the mixed HB/XB rotaxanes 3⋅XB-HB^DEG and 3⋅XB-HB^TEG, the signal arising from the internal benzene proton H_c is obscured by the stopper protons and could not be monitored, but it is assumed that anion binding similarly occurs via the macrocycle bidentate XB donor motif. Significant downfield movements of the axle triazole protons H_trz were also observed, suggesting participation of the triazole proton in anion binding via the formation of C−H⋅⋅⋅X⁻ HB interactions.

The perturbations of the selected proton signals (see Supporting Information for details) were monitored and used to determine 1 : 1 stoichiometric anion association constants for each rotaxane (Table 3). The neutral [2]rotaxanes display weak anion binding in 1 : 1 CD₃CN/CDCl₃, with all K_a values <200 M⁻¹. In particular, the halide association constants of 3⋅XB-XB^DEG and 3⋅XB-HB^DEG are significantly lower than the parent macrocycle 2⋅XB^DEG, which is presumably due to mechanical bond induced binding site inaccessibility. In contrast, the anion association constants of 3⋅XB-XB^TEG and 3⋅XB-HB^TEG remain comparable to the parent macrocycles, consistent with the reduced crowding of the larger TEG-based macrocycles.

Table 3. Anion binding constants (K_a/M⁻¹) of macrocyclic and interlocked receptors determined by ¹H NMR.^[a]

Guest	3⋅XB-XB^DEG	3⋅XB-HB^DEG	3⋅XB-XB^TEG	3⋅XB-HB^TEG
Cl^−	N.B.^[b]	40(1)	73(1)	62(1)
	M: 125(2)^[c] C: 66(1) ^[c]		M: 65(1)^[c] C: 84(1) ^[c]
Br^−	N.B.^[b]	77(1)	108(3)	91(2)
	M: 154(2)^[c] C: 78(1) ^[c]		M: 97(2)^[c] C: 142(1) ^[c]
I^−	N.B.^[b]	92(2)	150(5)	133(2)
	M: 186(3)^[c] C: 87(1) ^[c]		M: 146(3)^[c] C: 135(2) ^[c]

• [a] K_a values calculated using Bindfit with a 1 : 1 host–guest binding model. Errors are shown in parentheses and are all <2 %. All anions added as TBA⁺ salts. Solvent=1 : 1 CDCl₃/CD₃CN. T=298 K. [Receptor]=1.0 mM. [b] N.B.=No binding. [c] The italicised values after “M:” and “C:” are the previously reported K_a values of the analogous macrocycle and [2]catenane for the same anion. The K_a values of the catenanes have been statistically corrected by a factor of 2 to account for the presence of two bis(iodotriazole)benzene XB donor motifs.

Having determined both the cation and anion binding behaviour of the neutral XB [2]rotaxanes, their ability to recognise alkali metal halides as ion pairs was investigated. We first sought to determine this by calculating the binding enhancement factors, wherein the apparent anion binding affinities were determined in the presence of 1 equivalent of the metal cation and compared to that of the free rotaxanes. Whilst this approach is a reasonable assumption for cases in which the cation affinity is quantitative (i. e. K_a>10⁵ M⁻¹), when cation affinity is modest, the initial solution contains a significant proportion of un-complexed cation and host. This leads to a multitude of possible equilibria which are summarised in Figure 4 and discussed in further detail in the Supporting Information. The presence of these parallel equilibria, coupled with the dramatic conformational changes in the rotaxanes which also occur upon ion recognition, made us cognisant of the potentially highly complex recognition-induced chemical shift perturbation profile.

Figure 4

Schematic showing the possible equilibria occurring in solution during the ion-pair titration. R = receptor, M⁺ = alkali metal cation, X⁻ = halide anion.

To investigate these effects on determining ion-pair affinity, 3⋅XB-XB^TEG was selected as the subject of the study as it exhibited the largest anion and cation affinities of the [2]rotaxane series. To this end, ¹H NMR ion-pair titration experiments were undertaken in the conventional fashion of measuring an enhancement factor, in this case determining halide anion affinity 3⋅XB-XB^TEG in the presence of 1 equivalent K⁺ and comparing this value to that in the absence of K⁺. In a typical experiment the chemical shift of the rotaxane's macrocyclic internal benzene proton H_c was monitored with increasing anion concentration which in all cases for the halide titration experiments universally exhibited a downfield perturbation. Bindfit⁴⁴ analysis of the generated ‘binding’ isotherms, employing a 1 : 1 stoichiometric host–guest binding model, determined apparent K_a values of 116(3) M⁻¹ (Cl⁻), 2655(101) M⁻¹ (Br⁻) and 1442(84) M⁻¹ (I⁻). Taking these values at face value would suggest that 3⋅XB-XB^TEG exhibits considerable affinity and selectivity for KBr ion-pairs over the other potassium halide salts investigated. Of particular note is the ability of 3⋅XB-XB^TEG to bind a KCl ion-pair, which is impressive given the high lattice enthalpy of KCl. However, a closer and more critical analysis of the ¹H NMR spectra in their entirety indicates that these conclusions are not justified. Examining first the case of the ‘KCl’ titration with TBACl and KBAr₄^F, whilst superficially the ion-pair binding induced perturbations of proton signal H_c, appear consistent with a ‘KCl’ binding mode i. e. a single process, unidirectional movement of proton signals etc., inspection of the other signals in the [2]rotaxane reveal a significantly different scenario. Specifically, monitoring the perturbations in H₄, a signal from the stopper group of the axle component, upon the addition of 0–1 equiv. of TBACl an upfield shift was seen. However, with TBACl>1 equiv. a downfield shift is observed. Comparison of these spectra to that of the free rotaxane and those obtained during the anion titration experiments reveal that the addition of TBACl causes K⁺ decomplexation through exogenous ion-pairing (Figure 5a) and once the pre-added equivalent of K⁺ becomes saturated with a stoichiometric quantity of chloride, any further perturbations observed are a result of chloride-only binding by 3⋅XB-XB^TEG and not ion-pair ‘KCl’ binding.

Figure 5

Overlaid ¹H NMR ion-pair titration spectra of 3⋅XB-XB^TEG in the presence of 1 equiv. KBAr₄^F upon addition of (a) TBACl, (b) TBABr and (c) TBACl (500 MHz, 298 K, 1 : 1 CDCl₃/CD₃CN). The cartoon representations above each set of stacked spectra indicate the dominant equilibria occurring over the course of each titration.

In the case of the analogous ion-pair ‘KBr’ titration a similarly complex binding scenario is evidenced (Figure 5b). As in the case for the ‘KCl’ titration experiment, observing perturbations in proton signal H_C would suggest a relatively simple binding situation. Similar to the ‘KCl’ titration, an initial downfield shift of xylene proton signals H_7/8 is noted, in this case up to the addition of 4 equiv. of TBABr to the K⁺ precomplexed rotaxane 3⋅XB-XB^TEG, after which an upfield shift is observed with bromide equivalents up to 10 eq, whilst H_C is consistently perturbated downfield. In this case, it is clear that during the course of the titration endotopic ‘KBr’ ion-pair binding occurs up to 4 equiv. of Br⁻, followed by exotopic ion-pairing, i. e. gradual formation of exogenous KBr ion pairs. This suggests that ion-pair binding to 3⋅XB-XB^TEG dominates at low concentrations of bromide, but salt recombination competes significantly with ion-pair binding in the presence of excess bromide. The influence of these competing equilibria on the overall movement of the proton signals will considerably compromise any quantitative information concerning ion-pair binding affinities.

In contrast, analysis of the ¹H NMR spectral changes upon TBAI addition suggests that 3⋅XB-XB^TEG is capable of binding KI as an ion-pair. The macrocyclic internal benzene proton H_c exhibits a dramatic downfield shift with increasing anion concentration, indicative of enhanced anion binding by the metal-rotaxane complex. However, crucially the ethylene glycol proton signals in the 3.0–3.3 ppm region initially undergo chemical shift perturbations consistent with increased cation complexation, but appear to reach saturation in the presence of excess TBAI. This is consistent with a complexation mode in which iodide binding to 3⋅XB-XB^TEG simultaneously enhances the rotaxane's K⁺ affinity, such that the rotaxane exists predominantly as the metal-rotaxane complex and no further changes in the ethylene glycol region are observed (Figure 5c).

Given the significant salt recombination occurring in the ion-pair titrations of the rotaxane-cation mixtures with chloride and bromide, only the apparent iodide association constants were selected for further analysis and comparative studies (Table 4). Despite being an underestimation of the true anion binding affinity of the metal-rotaxane complexes, the determined apparent iodide binding constants of 3⋅XB-XB^TEG are significantly higher (>8-fold) than the neutral rotaxanes. This positive cooperativity between the cation and anion binding events is consistent with enhanced electrostatic attraction between the oppositely charged co-bound ions, similar to that of the previously reported [2]catenane systems.

Table 4. Selected apparent iodide association constants (K_app(I⁻)/M⁻¹) of [2]rotaxanes and [2]catenanes determined by ¹H NMR ion-pair titrations.^[a]

		[2]Rotaxanes		Analogous [2]Catenanes
Receptor	Cation	Proportion bound (%)^[b]	Kapp(I^−) (M^−1)	Proportion bound (%)	Kapp(I^−) (M^−1)
3⋅XB-XB^TEG	Na^+	24	1283(52)	33	818
	K^+	33	1442(84)	47	1050
3⋅XB-HB^TEG	Na^+	17	1053(40)^[c]	–	–
	K^+	21	844(36)^[c]	–	–

• [a] K_app values calculated using Bindfit with a 1 : 1 host–guest binding model. Errors (±) are shown in parentheses and are all <10 %. All anions added as TBA⁺ salts. Solvent=1 : 1 CDCl₃/CD₃CN. T=298 K. [Receptor]=[MBAr₄^F]=1.0 mM. [b] Initial proportion of metal-host complex calculated from cation association constants. [c] Estimated upper limit of K_app values for 3⋅XB-HB^TEG obtained by global fitting to H_trz and H_f (discussed in Supporting Information).

Surprisingly, despite the weaker cation binding affinities of the free rotaxanes, the cooperative ion-pairing effect appears to be more pronounced in the all-XB [2]rotaxanes than their catenane counterparts. 3⋅XB-XB^TEG exhibited stronger binding to iodide in the presence of either 1 equiv. Na⁺ or K⁺ than the analogous catenane (Table 4), despite the latter possessing higher association constants for both cations as well as possessing two bidentate bis(iodotriazole)benzene anion binding motifs. The higher degree of positive cooperativity in the ion-pair binding behaviour of 3⋅XB-XB^TEG relative to the TEG-based [2]catenane is tentatively attributed to the greater conformational flexibility of the rotaxane topology, which may allow the iodotriazole donors in both the macrocycle and axle to participate in cooperative anion binding.

To verify this hypothesis, attempts to evaluate the contribution of the axle iodotriazole groups to the ion-pair binding properties of the rotaxanes were undertaken by comparing the apparent anion binding constants of the all-XB and mixed XB-HB rotaxanes. Unfortunately, due to the significantly weaker cation association constants of the mixed XB/HB rotaxanes, the proportion of metal-complexed rotaxane initially present in solution is significantly lower, affecting the validity of such comparisons. In addition, the signal corresponding to H_c in 3⋅XB-HB^TEG is obscured by the stopper proton peaks and the sensitivity of the axle triazole proton H_trz to both cation and anion binding processes makes it challenging to select a suitable proton signal to monitor. This precludes a reliable determination of apparent anion association constants of metal-complexed 3⋅XB-HB^TEG, but the upper limits of the anion binding constants were estimated (see Supporting Information for full discussion) and found indeed to be lower than 3⋅XB-XB^TEG in all cases (Table 4).

Conclusion

A series of heteroditopic halogen bonding [2]rotaxanes for alkali metal halide ion-pair binding were constructed via alkali metal template-directed or active metal template-directed strategies. Extensive quantitative ¹H NMR cation, anion and ion-pair titration experiments were conducted on the rotaxane series. The co-conformations of the rotaxanes in the absence and presence of charged guests were investigated by ¹H NMR studies, revealing translocation of the macrocycle towards the central ethylene glycol stations upon alkali metal cation complexation to enable convergent metal coordination by the polyether regions of the macrocycle and the axle components whilst the halide anion recognition event results in minimal co-conformational change. Ion-pair affinity measurements were also undertaken, however, detailed analysis of the chemical shift perturbations revealed complex multiple parallel and competing equilibria, in addition to co-conformational changes, which severely complicated quantitative analysis of the rotaxanes’ ion-pair binding affinities. Importantly, comparing the apparent alkali metal cation iodide ion-pair binding K_app values of the [2]rotaxanes and the [2]catenane analogues revealed a notably higher degree of positive cooperativity between cation and anion binding events in the rotaxane topology, highlighting the advantages of greater mechanical bond co-conformational flexibility in the design of future multi-topic interlocked receptors.

Acknowledgments

H.M.T thanks the Clarendon Fund and the Oxford Australia Scholarships Fund for a scholarship. Y. C. T. thanks the Croucher Foundation for a scholarship. A.D. thanks the EPSRC for studentship funding (Grant reference number: EP/N509711/1.).

Conflict of interest

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.