Mechanical Bond Enhanced Lithium Halide Ion-Pair Binding by Halogen Bonding Heteroditopic Rotaxanes**
A previous version of this manuscript has been deposited on a preprint server (https://doi.org/10.26434/chemrxiv-2022-dzpcg).
Graphical Abstract
The first examples of selective lithium halide recognition by a mechanical bonded heteroditopic host is achieved using a series of halogen and hydrogen bonding [2]rotaxanes. A co-bound lithium cation switches on halide anion binding which increases greatly with increasing number of halogen bond donors. Ion-pair binding was not observed with the corresponding heteroditopic macrocycles, highlighting the importance of the mechanical bond effect.
Abstract
A family of novel halogen bonding (XB) and hydrogen bonding (HB) heteroditopic [2]rotaxane host systems constructed by active metal template (AMT) methodology, were studied for their ability to cooperatively recognise lithium halide (LiX) ion-pairs. 1H NMR ion-pair titration experiments in CD3CN:CDCl3 solvent mixtures revealed a notable “switch-on“ of halide anion binding in the presence of a co-bound lithium cation, with rotaxane hosts demonstrating selectivity for LiBr over LiI. The strength of halide binding was shown to greatly increase with increasing number of halogen bond donors integrated into the interlocked cavity, where an all-XB rotaxane was found to be the most potent host for LiBr. DFT calculations corroborated these findings, determining the mode of LiX ion-pair binding. Notably, ion-pair binding was not observed with the corresponding XB/HB macrocycles alone, highlighting the cooperative, heteroditopic, rotaxane axle-macrocycle component mechanical bond effect as an efficient strategy for ion-pair recognition in general.
Introduction
The fundamental roles of cation and anion species in a range of biological, industrial and environmental processes is the stimulus for the ever increasing demand to design and develop selective abiotic receptors.1-5 Due to their unique topologies and multidentate three-dimensional binding cavities, mechanically interlocked molecules6 (MIMs) have become increasingly popular as hosts for charged guest molecular recognition applications. In particular, the interlocked preorganised solvophobic binding sites can provide shape and size complementarity for a wide range of cationic,7-13 and anionic14-17 substrates. Importantly, MIMs often display significantly enhanced selective guest binding properties compared to analogous non-interlocked receptor counterparts.18 Heteroditopic receptors containing separate binding sites for both cations and anions, commonly exhibit augmented cooperative recognition behaviour relative to their monotopic receptor analogues.19, 20 This is generally attributed to favourable electrostatic interactions and/or conformational effects upon proximal ion-pair binding. Although the complexation of ion-pairs by non-interlocked heteroditopic receptors is now well established, mechanically bonded heteroditopic host systems for ion-pair recognition remain scarce.21-25 Furthermore halogen bonding (XB) heteroditopic interlocked receptors are extremely rare,22 despite the fact that XB interactions in general have shown superior anion guest binding properties compared to hydrogen bonding (HB) analogues.26
During the past two decades the worldwide demand for lithium has increased substantially due to its many applications, including lithium ion-batteries, pharmaceuticals and modern materials.27 However, despite its ubiquity in modern life, lithium salts can cause severe toxic effects due to their impact on the central nervous system. This includes irreversible neurological impacts such as dementia, Parkinson's disease, cerebellar impairment and peripheral neuropathies.28, 29 Recent reports show an alarming increase in lithium concentration in streams and tap water which corelates with the population density of the area.30 Therefore, the recognition of lithium salts is of great importance to minimise their environmental and biological effects. However, examples of heteroditopic receptors strategically designed to recognise LiX ion-pair species are rare and have been restricted to acyclic and macrocyclic ditopic receptors.26, 31-33
Herein, we report the synthesis and ion-pair binding properties of a series of novel heteroditopic [2]rotaxanes capable of lithium salt ion-pair recognition, through triazole based XB and HB interactions in both the axle and the macrocycle components (Figure 1). Binding of Li+ switches on the halide anion binding affinity of the rotaxane hosts in a cooperative manner through preorganised axle component-separated allosteric effects and charge assisted bonding interactions. Importantly, the MIM hosts displayed an enhanced ion pair binding mechanical bond effect when compared to their constituent macrocycles which increases with the number of XB donors in the interlocked host.
Schematic representation of the ion-pair bound rotaxane.
Synthesis of XB and HB rotaxanes
The heteroditopic macrocycle components, 5 and 6, of the target rotaxanes were designed to contain a 2,6 bis-methoxy pyridine motif in the first place to facilitate subsequent active metal template (AMT) MIM synthesis by endotopic copper(I) coordination.34 In addition, the respective macrocycle's pyridyl motif has the potential to coordinate the lithium cation. The macrocycles were prepared according to the multistep synthetic procedure shown in Scheme 1a. Bis-bromomethyl pyridine 2 was reacted with two equivalents of azide functionalized benzyl alcohol 135 in the presence of strong base NaH to form bis-azide precursor 3 in 70 % yield. Subsequent copper(I)-catalyzed azide–alkyne cycloaddition (CuAAC) ‘click’ reaction between the bis-azide 3 and the bis-alkyne 4 under high dilution conditions in anhydrous DCM afforded the target macrocycles 5 and 6 in 27 % and 29 % yields respectively. (Supporting Information, Section S1.2)
Synthesis of a) macrocycle 5 and 6, b) heteroditopic rotaxanes 9, 10, and 11.* 3.8 equiv. of Cu(MeCN)4PF6, at 30 °C for 5 days.
The synthesis of the target heteroditopic [2]rotaxane host systems was achieved via CuAAC active metal template (AMT)34 synthesis shown in Scheme 1b. To this end, macrocycles 5 and 6 were precomplexed with 0.95 equivalents of [Cu(MeCN)4]PF6 in CH2Cl2 and reacted with 5 equivalents of stopper precursors 734, 36 and 834 in the presence of catalytic amounts of DIPEA at room temperature for 48 h to afford rotaxanes 9 and 10 in 34 % and 35 % isolated yields respectively. The synthesis of the all XB rotaxane 11 proved much more challenging, presumably due to steric effects by additional XB donor atoms hindering the formation of the interlocked structure. However, by increasing the reaction time to 5 days, temperature of the reaction to 30 °C and adding more [Cu(MeCN)4]PF6 catalyst to the reaction mixture, rotaxane 11 was successfully isolated following purification in a modest yield of 9 %. Full synthetic details and procedures can be found in the Supporting Information.
All novel structures were characterized by 1H NMR and 13C NMR spectroscopy, and high-resolution electrospray ionisation (ESI) mass spectrometry. Evidence for the successful formation of the interlocked structures 9,10 and 11 was shown by comparison of 1H NMR spectra of the rotaxane product and non-interlocked macrocycle and axle components. In particular, diagnostic upfield shifts of the macrocyclic hydroquinone protons H5 in the interlocked product were indicative of the aromatic donor-acceptor nature of the electron rich macrocyclic hydroquinone groups and electron deficient triazole axle.
Further evidence for the interlocked nature of the synthesised compounds was provided by 2D ROESY NMR spectroscopy (see Supporting Information, Section S1.3).
Solid state structure
Crystals of the XB macrocycle hydrochloride 6.HCl suitable for single crystal X-ray diffraction structural analysis were obtained by slow evaporation of a CDCl3 solution of the macrocycle in the presence of excess of TBACl (Figure 2).37 The solid state structure reveals the two iodo-triazole based XB donors to exhibit near linear (average 170°) XB interactions with the Cl− anion in a concerted manner. A significantly shorter distance between the XB donors and Cl− compared to their van der Waals radii (average 87 %) is also observed. The macrocycle's pyridyl motif was observed to be protonated, presumably due to trace acid residues in CDCl3. The structure thus highlights the respective cation and anion binding sites of the heteroditopic macrocycle.
Crystal structure of XB macrocycle hydrochloride 6.HCl. Colour code of atoms: O-red, N-blue, I-purple, H+-white, Cl−-green, C-teal. Non-covalent host-guest interactions are shown in yellow.
Anion and Ion-pair binding properties of rotaxanes and macrocycles
In order to establish the lithium halide ion-pair binding mode of the heteroditopic [2]rotaxanes 9, 10, 11 and macrocycles 5 and 6, preliminary qualitative 1H NMR ion-pair complexation experiments were performed. In a typical experiment, the rotaxane was complexed with one equivalent of LiClO4 followed by sequential addition of halide anions (as their TBA salts) in CD3CN:CDCl3 solvent mixtures. This particular solvent combination was chosen based on the solubility of host and potential guest ions.
All rotaxanes exhibited similar 1H NMR perturbations upon the initial addition of the lithium cation and subsequent halide anion (See Supporting Information, Section S3.2 and S4.2). A truncated 1H NMR spectrum of rotaxane 10 with the addition of LiClO4 and TBABr is shown in Figure 3. Upon addition of LiClO4, significant perturbations of the rotaxane protons were observed in the axle methylene signal He and terphenyl stopper signal Hf, as well as the macrocyclic methylene signal H7 adjacent to the pyridyl moiety, indicating Li+ complexation in the predicted cation binding cavity. Proton H3 in the anion binding site did not show any perturbations indicating no coordination of the perchlorate counteranion from the Li salt. Subsequent titration of TBABr caused prominent perturbations to proton signals H2, H3, Hf and Ha, which are local to the postulated anion binding site. Additionally, all-HB rotaxane 9 displayed prominent peak perturbations of triazole protons H10 and Hg upon the addition of anions, further indicating the direct involvement of both macrocycle and axle triazole C−H donors in the anion recognition process (Figure S3.1). While not observed upon titrations with TBABr and TBAI, the addition of TBACl to lithium precomplexed rotaxanes caused decomplexation of the cation in all cases, as evident by the observation of free rotaxane species in 1H NMR spectra upon addition of the Cl− salt (Figure S3.1a).
Truncated 1H NMR spectra in 1 : 9, CD3CN:CDCl3, of a)Rotaxane 10 b) Rotaxane 10 in the presence of 1 equivalent of LiClO4 c) Rotaxane 10 in the presence of 1 equivalent of LiClO4 and TBABr.
This may be attributed to the strong lattice enthalpy of LiCl driving salt recombination.38
Quantitative ion-pair binding data was determined by monitoring non overlapping 1H NMR chemical shifts of rotaxane protons H2,H5,Hf and Ha,H10 in the presence and absence of one equivalent of LiClO4 as a function of halide anion concentration. Titration isotherms generated by monitoring protons H2 and H10 of lithium precomplexed rotaxane with the addition of halide anions are shown in Figure 4. With the lithium complexed rotaxanes, Bindfit39 analysis determined 1 : 1 stochiometric host-guest association constants displayed in Table 1.40 All rotaxanes precomplexed with lithium were found to bind Br− and I− anions in CD3CN:CDCl3 solvent mixtures. However, no anion binding was observed in the absence of Li+. The presence of the bound metal cation in the rotaxane host effectively ‘switches on’ the anion binding capability, due to favourable proximal electrostatic interactions between the two co-bound ions and the preorganisation of the anion binding site via the mechanical bond (see DFT calculations below).
Anion binding isotherms generated by monitoring perturbations of the H10 triazole proton signal of lithium precomplexed rotaxane 9 and H2 internal tert-butyl benzene proton signal of lithium precomplexed rotaxane 10 and 11 upon addition of TBABr and TBAI, 298 K, 500 MHz; a) in 1 : 9 CD3CN:CDCl3 solvent mixture, b) in 3 : 7 CD3CN:CDCl3 solvent mixture.
Anion |
Cation |
M.5 |
M.6 |
R.9 |
R.10 |
R.10 |
R.11 |
|---|---|---|---|---|---|---|---|
|
|
1 : 9 CD3CN:CDCl3 |
3 : 7 CD3CN:CDCl3 |
||||
Cl− |
Li+ |
– |
– |
[b] |
[b] |
– |
[b] |
Br− |
Li+ |
[b] |
[b] |
6946(9) |
>104 |
5666(7) |
>104 |
I− |
Li+ |
[b] |
[b] |
2369(7) |
>104 |
2890(5) |
3419(7) |
- [a] Ka values were calculated using the global fit option in Bindfit software using 1 : 1 binding model. Errors (%) are in parenthesis. Li+ is added as LiClO4 and all anions as their TBA salts [Receptor]=1 mM, T=298 K. [b] Salt recombination
Interestingly, Table 1 highlights the increased strength of halide anion binding in these interlocked hosts as the number of XB sigma-hole donors in the cavity increase. In 1 : 9 CD3CN:CDCl3 mixtures, a significant increase in the magnitudes of association constants of LiBr and LiI is observed for the mixed XB-HB rotaxane 10 compared to its all HB rotaxane analogue 9. As the Ka values for 10 were >104 M−1, a more competitive solvent mixture system of 3 : 7 CD3CN : CDCl3 was used to obtain Ka values for rotaxanes 10 and 11. Replacement of the axle's single triazole HB bond donor of 10 with an XB bond donor in 11 resulted in a large enhancement of halide anion binding strength, particularly with the smaller Br− anion. All the rotaxane hosts showed a two- to three-fold selectivity for the LiBr ion-pair over LiI. This may be attributed to the relative higher basicity of Br− and the size complementarity of LiBr to the preorganised ion-pair binding cavities of the rotaxanes.
To further investigate the selectivity of rotaxane 10 for anions, analogous 1H NMR titrations were performed for the lithium precomplexed rotaxane (10.Li) with the sequential addition of AcO−, NO3− and SCN− anions. In all instances, Li+ decomplexation was observed upon addition of the respective anion (Figure S3.3), even though the lattice energies of LiNO3 and LiSCN are lower than the lattice energy of LiBr.38, 41 Hence, the selectivity presumably stems from the preferential halide binding properties of rotaxane 10, specifically the binding of the softer Br− and I− anions to XB donors42 and size complementarity of halide anions to the anion binding site.
Attention then turned to investigate the selectivity of rotaxane 10 for group I metal cations. Due to poor solubility of the receptor in the presence of NaClO4, NaBArF was used as the Na+ salt. 1H NMR titration experiments in 3 : 7 CD3CN:CDCl3 determined the 1 : 1 stoichiometric association constant for Li+ (Ka=697(1) M−1) was more than three times higher than the association constant value for Na+ (Ka=181(1) M−1) (See Supporting Information, Section S3.3). This may be attributed to size complementarity and higher charge density of Li+ cation.
In contrast to the rotaxanes, HB macrocycle 5 and XB macrocycle 6 did not bind any LiX ion pairs in 1 : 9 CD3CN:CDCl3. Upon addition of LiClO4, the proton signals adjacent to the respective macrocycle's pyridyl moiety, H6 and H7 shifted downfield, and H8 shifted upfield. This suggests complexation of Li+ in the proximity of the pyridyl cation binding site, as observed in the solid state structure (Figure 2). Upon subsequent addition of TBABr or TBAI, these signals reverted back to their original metal cation free chemical shift values, suggesting potential ion-pair formation outside of the respective macrocycle, preventing effective ion-pair recognition (Supporting Information, Section S3.2).
DFT calculations
Density functional theory (DFT) studies were performed to investigate the binding modes of lithium halides to rotaxanes 9,10,11 and macrocycles 5,6. Simulations were carried out in the gas phase using Gaussian 1643 at the B3LYP level of theory44 using aug-cc-pVTZ basis set for chlorine atoms, aug-cc-pVTZ-PP basis set for larger halide atoms, and 6–31 g* basis for other atoms.45-47 Counterpoise correction was invoked to correct for the basis set superposition error.48, 49 The initial input structures were generated from modifying solid state crystal structures of macrocycle 6 and of the HB triazole axle.50 The terminal terphenyl stoppers of the axle were replaced by methyl groups for ease of simulation.
A geometry optimised structure of LiBr bound rotaxane 11 is shown in Figure 5. According to the optimised structure, Li+ interacts with the pyridine nitrogen and the adjacent oxygen of the macrocycle, and the triazole nitrogen and the neighbouring oxygen of the axle, in a four-coordination geometry. The bromide anion interacts with all three triazole XB bonds of both the axle and the macrocycle, forming characteristic near linear interactions (1680–1720). The average C−I-X− bond length of bromide bound rotaxanes 10 and 11 is 82.5 % of the sum of the bromide ionic radius and the iodine van der Waals radius. The corresponding value for iodide bound rotaxanes 10 and 11 is 85 %, which is in line with the experimental evidence of stronger halogen bonding for bromide bound rotaxanes. The optimised structures agree with the corresponding solution 1H NMR titration data where significant proton perturbations were observed in the postulated anion and cation binding sites (Supporting Information, section S4.2).
DFT optimised structure of LiBr bound rotaxane 10. H atoms are omitted for clarity. Colour code of atoms: O-red, N-blue, I-purple, Li-pink, Br- brown, C(axle) -gray C(MC)-teal. Non-covalent host-guest interactions are shown in yellow.
Similarly, DFT calculations of HB macrocycle 5 and XB macrocycle 6 show that Li+ interacts with pyridine nitrogen and the adjacent oxygen atoms (Supporting Information, Section S4.1). This is confirmed by solution 1H NMR titration data where peaks H8 and H7 of the macrocycle are shifted significantly, upon addition of the cation (Supporting Information, Section S4.2).
Calculated lithium halide complexation energies for all rotaxanes are presented in Table S4. Aligning with 1H NMR titration results, complexation energies of the rotaxanes were far superior compared to their analogous macrocycles. For all rotaxanes LiBr showed a slight increase of complexation energy compared to LiI. It is however important to note that solvophobic effects from XB donors are not accounted for in gas phase calculations. Therefore, it is difficult to compare the small changes in association constants of rotaxane 9, 10, 11.
The second sphere electrostatic effect of Li+ for cooperative binding of anions is reflected in the distribution of electrostatic potential mapped onto the electron density surface. The isoelectric maps for rotaxanes 10 and 11 are presented in Supporting Information, Section S4.4. Binding of Li+ increases the potential energy of the proximal XB/HB donor atom of the axle. This is evident from the increase of potential energy surrounding the axle iodine atom of rotaxane 11 from 245 kcal mol−1 to 301 kcal mol−1, and axle H atom of rotaxane 10 from 195 kcal mol−1 to 257 kcal mol−1 when Li+ is bound.
Conclusions
A family of XB and HB heteroditopic [2]rotaxane hosts with axle separated binding cavities for simultaneous cation and anion binding were synthesised for lithium halide ion pair recognition. 1H NMR titration studies and DFT calculations revealed the interlocked rotaxane hosts to show superior ion-pair binding for lithium halides compared to their non-interlocked macrocycle components. Binding of Li+ was found to both polarise the XB donors and preorganise the anion binding site, leading to a cation-induced ‘switch on’ of cooperative halide recognition. Quantitative ion-pair binding data revealed all rotaxanes to display notable selectivity for LiBr over LiI. In addition, the strength of bromide and iodide halide binding was demonstrated to increase significantly with increasing number of XB donors in the interlocked rotaxane cavity, making all XB rotaxane 11 the strongest ion-pair binding rotaxane of the family. Importantly, these results serve to highlight how the unique cooperative rotaxane axle-macrocycle component mechanical bond effect, in combination with powerful XB sigma-hole interactions, creates highly effective heteroditopic ion-pair host systems. Exploiting mechanical bond design for MIM heteroditopic construction as an efficient strategy for ion-pair recognition is continuing in our laboratories.
Acknowledgements
V.K.M, D.M and J.P contributed equally. V.K.M thanks the E.P.A Cephalosporin Scholarship fund, Linacre College for financial support. D.M thanks the EPSRC Centre for Doctoral Training, Theory and Modelling in Chemical Sciences, under Grant No. EP/L015722/1 and Linacre College for a Carolyn and Franco Giantruco Scholarship and the Department of Chemistry, University of Oxford for financial support. J.P thanks the EPSRC for postdoctoral funding (EPSRC Grant No. EP/P033490/1). Authors would like to acknowledge the use of the University of Oxford Advanced Research Computing (ARC) facility in carrying out this work.
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.
