Chiral hemicucurbit[8]uril as an anion receptor: selectivity to size, shape and charge distribution

Chiral (all-R)-cyclohexanohemicucurbit[8]uril binds anions in a 1 : 1 ratio in pure methanol like a molecular Pac-Man™ with remarkable selectivity based on the size, shape and charge distribution of the anion.


Mass Spectrometry ESI-MS experimental
Mass spectrometry experiments were performed with AB Sciex QSTAR Elite ESI-Q-TOF mass spectrometer, equipped with an API 200 TurboIonSpray ESI source from AB Sciex. Nitrogen was used as drying and nebulization gas. The parameters were optimized to get the maximum abundance of ions under study. The measurements and data handling was done with Analyst ® QS 2.0 software. The ions were characterized according to comparison of theoretical and experimental m/z values and isotopic patterns calculated on basis of natural abundances of elements.
Samples were prepared using HPLC solvents. From host 1 mM stock solution was made in methanol. From salts 2 or 5 mM stock solutions were prepared in methanol. All samples were prepared with 10 µM concentration and 1:1 host:guest ratio in methanol. The samples were injected into the ESI source with 5 µl flow rate and spectra were externally calibrated using sodium trifluoroacetate.
Competition experiments were performed with 1:1:1 ratio using two competing anions with same counter cation (Na + or TBA + ) at the time. Competition experiments were carried out on five different samples, which all were measured five times. The overall variance was calculated from the standard deviation of sampling and the standard deviation of the measurement (s tot = s 1 + s 2 ). Measurements or samples were rejected if the average deviation of a suspect value from the mean was four or more times the average deviation of the retained values.
In CID experiments the ions were isolated and activated by CE values from 5 to 55 V. Nitrogen was used as a collision gas with 5 bar pressure. The dissociation was followed as a function of CE value and dissociation graphs were fitted to sigmoidal. For studied ions CE 50% values were defined using OriginPro 2015 software.

Complexation and competition experiments
Complexation of cycHC [8] was monitored on (-)ESI-MS mode in methanol, where only free anion and 1:1 host:guest complexes were observed. Based on complexation experiments, the affinity of the host was studied in competition experiments, where two anions with the same counter cation were added to the same sample. The order in affinity was determined as follows: SbF6 − ≈ PF6 − > ReO4 − > ClO4 − > SCN − > BF4 − > HSO4 − > CF3SO3 − > H2PO4 − > AcO − . The competition between SbF6 − and PF6 − was performed several times, but no clear preference was observed.
Tetrahedral anions with smaller packing coefficient seem to form kinetically more stable complexes with cycHC [8]. This is most probably due to size fit into the cavity. While anions with smaller packing coefficient have void space around them, higher kinetic activation is needed for these complexes to dissociate. Respectively, guests with higher packing coefficient are more sensitive for thermal activation, and lack of void space inside the cavity results in low-energy dissociation.

Crystallographic details for TBP(SbF6@cycHC[8]) • 2CH3OH
Due to the commercial unavailability of tetrabutylammonium or tetrabutylphosphonium hexafluoroantimonate salts, an amount of TBPSbF6 sufficient for threefold excess over cycHC [8] was prepared ahead of crystallization. For this equimolar amounts of NaSbF6 and TBPBr were weighed out and dissolved separately in distilled water. Upon mixing the two aqueous solutions a white precipitate formed. The precipitate was extracted from the suspension by repeated washing with by CH2Cl2. The organic phase was collected and subsequently dried, resulting in a white solid. Neither the purity nor the yield of the product was determined. The solid was dissolved in methanol and added to cycHC [8]. The solution was sonicated and heated for all of the cycHC [8] to dissolve. When the clear solution was left to cool down, small colourless block-like single crystals of TBP(SbF6@cycHC [8]) • 2CH3OH formed. The compound was found to crystallize in monoclinic space group P21, with unit cell dimensions very close to orthorhombic [β = 90.0578°]. The crystal exhibited twinning by pseudo-merohedry, imitating orthorhombic Laue symmetry. Twin law, a two-fold rotation around the crystallographic c-axis (−100 0−10 001), was found using the TwinRotMat tool of PLATON 10 and applied in the refinement. The fraction of the minor twin component refined to 0.2056 (6). The unit cell contains four TBP(SbF6@cycHC [8]) moieties together with 8 solvent molecules.

) • 2CH3OH
Single crystals of the complex were obtained from a methanol solution of cycHC [8] with about twofold excess of TBABF4 upon slow evaporation of the solvent. The inclusion complex crystallized in an orthorhombic space group P212121, with four units of TBA(BF4@cycHC [8]) together with eight solvent molecules in the unit cell. The tetrafluoroborate anion was refined in four separate disorder components, with sof of their respective relative occupancies refined freely. The resulting relative sof of the disorder components was 0.352(7)/0.163(7)/0.251(8)/0.23 (8). The geometry of the disordered BF4 − anions was restrained to be tetrahedral, with tying the 1,2-and 1,3-distances (BF and FBF respectively) to a free variable which was then refined. One of the two methanol molecules was modelled as occupying two discreet positions, with relative occupancies 0.601(6)/ 0.399 (6).

) • 2CH3OH
Single crystals of the complex were obtained from a methanol solution of cycHC [8] with an equimolar ratio of TBAClO4 upon slow evaporation of the solvent. The inclusion complex crystallized in an orthorhombic space group P212121, with four units of TBA(ClO4@cycHC [8]) together with eight solvent molecules in the unit cell. The perchlorate anion was refined in four separate disorder components, with sof of their respective relative occupancies refined freely. The resulting sof of the disorder components was 0.291(8)/0.219(8)/0.252(8)/0.237 (8). The geometry of the disordered ClO4 − was restrained to be tetrahedral, with the 1,2-and 1,3-distances (CO and OCO respectively) tied to a free variable, that was then allowed to refine. The disorder of one of the butyl groups of TBA was modelled between two discreet positions, with sof or the respective disorder components refined to 0.549(12)/0.451 (12). One of the two methanol molecules in the asymmetric unit was modelled as occupying two discreet positions, with relative occupancies 0.857(7)/0.143 (7).

Crystallographic details for TBP(IO4@cycHC[8]) • 2CH3OH
An amount of TBPIO4 sufficient for threefold excess over cycHC [8] was prepared ahead of crystallization. For this equimolar amounts of NaIO4 and TBPBr were weighed out and dissolved separately in distilled water. Upon mixing the two aqueous solutions a white precipitate formed. The precipitate was extracted from the suspension by repeated washing with by CH2Cl2. The organic phase was collected and subsequently dried, resulting in a white solid. Neither the purity nor the yield of the product was determined. The solid was dissolved in methanol and added to cycHC [8]. The solution was sonicated and heated for all of the cycHC [8] to dissolve. When the clear solution was left to cool down, small colourless block-like single crystals of TBP(IO4@cycHC [8]) • 2CH3OH formed.
The compound was found to crystallize in monoclinic space group P21, with unit cell dimensions very close to orthorhombic [β = 90.0371 (12)°]. The crystal exhibited twinning by pseudo-merohedry, imitating orthorhombic Laue symmetry. Twin law, a two-fold rotation around the crystallographic c-axis (−100 0−10 001) was found using the TwinRotMat tool of PLATON 10 and applied in the refinement. The fraction of the minor twin component refined to 0.3228 (6). The unit cell contains four TBP(IO4@cycHC [8]) moieties together with 8 solvent molecules. The two metaperiodide anions in the asymmetric unit were modelled in three disorder components each, with the occupancies of the respective disorder components refined freely, giving sof 0.431(3)/0.377(3)/0.192(3) and 0.346(3)/0.530(2)/0.123 (2) respectively for components in the two anion sites. The geometry of the disorder components of IO4 − was restrained to be tetrahedral, with the 1,2-and 1,3-distances (IO and OIO respectively) tied to a free variable, which was then allowed to refine. One of the four methanol molecules in the asymmetric unit was modelled as occupying two discreet positions, with relative occupancies tied to a free variable which was then refined, giving sof 0.708(6)/0.292 (6) for the disorder components.

) • 2CH3OH
Single crystals of TBA(ReO4@cycHC [8]) • 2CH3OH were obtained from an equimolar solution of cycHC [8] and TBAReO4 in methanol by slow evaporation of the solvent. The compound was found to crystallize in space group P212121, with four TBA(ReO4@cycHC [8]) moieties in the unit cell together with 8 methanol molecules. The perrhenate anion was modelled in two disorder components, with the occupancy of the disorder components refined freely. The resulting relative sof of the disordered anion components was 0.597(4)/0.403 (4). The geometry of the ReO4 − was restrained to be tetrahedral, with the 1,2-and 1,3-distances (ReO and OReO respectively) tied to a free variable, which was then allowed to refine. One of the two methanol molecules in the asymmetric unit was modelled as disordered between two discreet orientations, with relative occupancies 0.64(5)/0.36 (5). The other methanol molecule was also found to be disordered, but the minor component could not me modelled properly. The major component in this solvent site was modelled with 0.75 occupancy.

) • 2CH3OH
Single crystals of TBA(CF3SO3@cycHC [8]) • 2CH3OH were obtained from an equimolar solution of cycHC [8] and TBACF3SO3 in methanol by slow evaporation of the solvent. The compound was found to crystallize in space group P212121, with four TBA(CF3SO3@cycHC [8]) moieties in the unit cell together with 8 methanol molecules. The encapsulated trifluoromethanesulfonate anion was modelled in two disorder components, crossed to one another at an angle of about 67°. The relative occupancy of these was refined freely, resulting in sof 0.876(3)/0.124 (3). The geometry of the CF3SO3 − was restrained to be more symmetrical and same for the two modelled disorder components. Two additional disorder components are suggested by looking at the anisotropic displacement parameters of the modelled CF3SO3 − . The ADPs of the sulphur atoms are unexpectedly large while the carbon atoms have relatively small ADPs, indicating that each of the CF3SO3 − orientations probably has an additional overlapping disorder component flipped 180 degrees respective to it. However attempts to refine these additional disorder components were not successful as the geometry of the flipped minor components was severely affected by their overlap with the major disorder components. One tetrabutyl group of the TBA + was seen to be disordered, and was therefore modelled in two disorder components. The relative occupancy refined to 0.563(12)/0.437 (12) for the disorder components. One of the two methanol molecules in the asymmetric unit was modelled as disordered between two discreet orientations, with relative occupancies 0.738(6)/0.262 (6). Restraints were applied to the geometry of the solvent molecules. Restraints were also applied to the ADPs of the disordered parts of the structure. Some reflections, obscured by the beam stop, were omitted from the refinement.

Analysis of Hirshfeld surfaces and host-guest interactions
Hirshfeld surfaces 11 provide a valuable tool for analysing the close contacts between moieties in the crystal structures. The surface is created around a molecule, where its electron density exceeds that from all the neighbouring molecules. In this work, the Hirshfeld surfaces were calculated for SbF6 − ( Figure S6), IO4 − ( Figure S7) and CF3SO3 − ( Figure S8), which represent the different shape of guest anions encapsulated. The program Crys-talExplorer 9 was used to generate the Hirshfeld surfaces and to highlight the close contacts between the anion and the neighbouring host molecule (C−H•••anion). The surfaces were mapped with dnorm where a red spot sig-  Shortest distances between the host molecule and the encapsulated anions were investigated using Mercury 3.7, 7 the result of which is shown in Tables S2 -S8. Distances shorter than the sum of van der Waals radii of hydrogen and the acceptor atom ∑r(vdW)[H, A] are marked in black, as they indicate influential host-guest interactions. The denomination of atoms in the host molecule is as follows: equivalent monomeric units are assigned a suffix A-H (P) depending on the number of monomers in the asymmetric unit, atoms in the monomeric units are labelled as shown on Figure S9. Anions are labelled with suffixes when more than one moiety of the host-guest complex is present in the asymmetric unit or/and when numbering disorder components in one anion site.  Figure S9. Atom labelling scheme shown on the monomeric unit of (all-R)-cycHC [8]. The placement of the model anion (light blue) indicates the side of the monomer facing the cavity. Numbers in red represent the numbering scheme of carbons.
Hexafluoroantimonate (SbF6 − )    Figure S13. Close contacts within the second moiety of IO4@cycHC [8] in the asymmetric unit, (monomer suffixes I-P). The contacts for the three disorder components are shown separately for clarity, but they occupy the same site with sof 0.346 (3)  ratio. Chemical shift of the host proton 2ax was followed, with the exception of NaSbF6, where due to the broadening of proton 2ax signal, proton 1ax was followed. The maximum shift change was seen at α = 0.5 in all of the Job plot curves, indicating a 1:1 molecular association ( Figure S17). Guest affinity was ascertained by 1 H NMR titration experiments in deuterated methanol at 288 K. The concentration of cycHC [8] for a titration experiment was chosen according to the predicted guest binding strength (host concentrations 4mM, 3mM, 0.4mM or 0.04mM were used) and held constant throughout the titration sequence. The guest stock solution was added to the titration samples in small aliquots cumulatively and the spectra recorded upon each addition. The concentration of guest stock solution was chosen so that the total volume of added guest stock solution would not affect the overall concentration of the components in the sample more than 10%. The 1 H NMR spectra were recorded on a Bruker Avance III 800MHz spectrometer using regular 5 mm NMR tubes. Probe temperature was set to 288 K and each sample was allowed to thermally equilibrate in the probe for at least 10 min prior to measurements. The progressive changes of the chemical shifts of 1ax, 2ax and 3eq protons were followed as small aliquots of guest stock solution were added to a solution containing cycHC [8], resulting in a set of spectra (18-21 per titration) for each of the tested guests. Precise host-guest ratio in the titration sequence was determined by integration of the 1 H signal for the host 1ax proton (2.91-2.82ppm) and either the methyl signal (1.03 ppm) of guest counter cation TBA or the added internal standard 1,3,5-tris(trifluoromethyl)benzene (8.3515 ppm). The relaxation time for the signals used for quantification was measured on concentrations relevant to the titration experiment setup, using the t1ur pulse programme. This gave relaxation times for: cycHC [8] proton 1ax 647 (8)  Where f is the measured chemical shift, y0 is the chemical shift of the empty cycHC [8], Dy is the difference of the chemical shift of an empty host and a fully occupied host, Ka is the association constant, P is the total host concentration and x is the total guest concentration Following three signals instead of only one afforded a reliable estimation of the association constant, since the data is less biased by the possible errors on reading the chemical shifts, especially since the migrating signals were seen to overlap at certain guest ratios in the titration sequences.   The pulse program of this measurement was not suitable for quantitative evaluation of the standard, 1,3,5-tris(trifluoromethyl)benzene (at 8.3515 ppm) signal intensity, as the signal was suppressed due to insufficient relaxation delay between scans (ideal P1=30 and D1= 5s, used pulse program P1=90 and D1=1s). The standard signals were therefore reassessed using a calibration curve obtained from measuring a number of the titration samples of cycHC [8]+NaPF6+Std at each of the pulse programs P1=90, D1=1s and P1=30, D1=5s and evaluating the increase of the standard signal relative to the corresponding 1ax signal of the cycHC [8]. Added guest concentration, used in the curve-fitting of the binding isotherm, was calculated based on the corrected integral values for the standard,

TBACF3CO2
TBACF3CO2 was prepared ahead of titration from trifluoroacetic acid ant tetrabutylammonium hydroxide. 600μL of 3mM cycHC [8] stock solution was used for each of 5 titration samples. A 0.36M (360mM) TBACF3CO2 stock solution was prepared and added to titration samples cumulatively. Precise determination of host-guest ratio in a titration sequence was determined by integration of the 1 H signal for host 1ax proton (2.91-2.82ppm) and the methyl group (1.03ppm) of guest counter cation TBA. Spectra with up to 15.23 eq of TBACF3CO2 were collected, and from these it was assessed that Ka must be lower than 10. Determination of the precise association constant was not conducted, as the overall shift was so small even in the case where large excess of TBACF3CO2 was added.

ITC measurements
Calorimetric measurements were performed on a MicroCal iTC200 calorimeter (GE Healthcare Life Sciences). The volume of the calorimetric cell was 200 µl and the size of the syringe was 40 µl. All experiments were carried out in methanol. In the case of the interaction between NaSbF6 and cycHC [8], calorimetric cell was loaded with 0.2 mM NaSbF6 solution and the syringe was filled with 2.4 mM cycHC [8] solution. Sequential injections (1 µl) of cycHC [8] were added at 60 s intervals by rotating syringe to the calorimetric cell. Since the heat of dilution was negligible, it was not subtracted from the total heat of the interaction. Such a protocol was chosen as the heat of dilution of cycHC [8] was negligible compared to the heat of dilution observed using the protocol of sequential addition of a concentrated NaSbF6 solution to the calorimetric cell containing a cycHC [8] solution.
Using the same protocol and solution concentrations was attempted for studying the interactions of NaPF6 with cycHC [8], but was deemed not optimal for obtaining a well-fitting binding isotherm due to the lower binding strength of PF6 − . Simple increasing of the solution concentrations while using the same protocol was not applicable as a solution of 24mM of cycHC [8] is outside the solubility limits of cycHC [8] in methanol. Thus in order to investigate the interaction between NaPF6 and cycHC [8], higher solution concentrations were achieved by loading the calorimetric cell with 2.5 mM cycHC [8] solution and filling the syringe with 31.39 mM NaPF6. Sequential injections (1 µl) of NaPF6 were added at 60 s intervals by rotating syringe to the calorimetric cell. The heat of dilution was obtained by injecting NaPF6 to methanol. Prior to data analysis the heat of dilution was subtracted from the corresponding total heat of the interaction. All measurements were carried out at 298 K and the stirrer speed was set to 600 rpm. ORIGIN 7.0 with MicroCal AddOn was used for data analysis. Data were fitted using simple one-site model.
The deviation from n = 1 could be caused by small uncertainties of the used solutions concentrations. In NMR spectroscopic methods the concentration of the solutions can be verified by an internal standard, whereas the concentrations of the solutions used in the ITC measurements rely on the preparation of the stock solutions. The uncertainties in concentration would reflect in the stoichiometry parameter, which is allowed to vary in the fitting of the binding isotherm. The resulting n = 0.82-0.86 is nevertheless close enough to n = 1, and 1:1 stoichiometry of binding is also supported by 1 H-NMR Job plot analysis in methanol, X-ray crystallography in solid state and MS in gas phase.
Variable temperature NMR for NaSbF6 and NaPF6 inclusion complexes

General methods
The 2ax and 3eq proton resonances for NaPF6 (1.5 mM) and 3eq proton resonances for NaSbF6 (1.5 mM) in the variable temperature NMR spectra of a methanol-d4 solution of cycHC [8] (2.6 mM) in the temperature region of 204295 K in 4 K temperature increments were simulated with the program WINDNMR V. 7.1.14 (http://www.chem.wisc.edu/areas/reich/plt/windnmr.htm) after processing the NMR spectra with NUTS. The coalescence temperatures were 241 K for NaPF6 and 253 K for NaSbF6, respectively. The 3eq and 2ax resonances for NaPF6 and the 3eq resonance for NaSbF6 were present as free and complexed cycHC [8] in the slow exchange region at low temperatures and were therefore suitable for simulations. DNMR option with 2dddd was used in WINDNMR to simulate spectra with all input parameters except temperature taken from the slowexchange or near slow-exchange limit spectra. Adjustments of spectral amplitudes and rate constants (kassoc + kdissoc) were carried out iteratively until the line-shapes matched. The coalescence region was neglected for determining activation parameters due to large broadening of the peaks and difficulty in obtaining good fits. The simulated spectra are shown in shown in Figure S20 and Figure S21 and fitting data are listed for NaSbF6 in Table S10 and for NaPF6 in Table S11. Temperatures were calibrated using an external standard MeOH solution.
NMR titration experiments and ITC data indicate that the equilibrium greatly favors the host-guest complex. Furthermore, it was not possible to detect either the free host or a precomplex with PF6 − at 0.4 mM concentration and in 1.5 times excess of the guest at sub-coalescence temperature of 213 K. This also suggests that the equilibrium constant for the overall reaction is greatly in favor of the host-guest complex. Accordingly, the dissociation rate constant is negligible compared to the association rate constant. Therefore, the observed sum of rate constants, obtained from lineshape analysis in WINDNMR, is approximately equal to the rate constant of the association reaction. Based on the assumption that kasssoc + kdissoc = kassoc, the calculated activation parameters, obtained from the Eyring analysis, are shown in Table 2 in the main text and the Eyring plots brought in Figure  S 19. Experimental details to confirm a first-order association reaction are given below.
Activation parameters for both the dissociation and association reactions were obtained from modified Eyring equation: via a linear plot ln(k/T) versus 1/T.
The activation Gibbs free energy, enthalpy and entropy are detailed in Table 2 in the main text.

Association-dissociation reaction order
The dissociation reaction is unambiguously a first-order process (H -host; G -guest, HG -complex):

HG ⟶ H + G
The dissociation reaction rate is given by the equation:

= ×
We have determined that the association reaction is also a first-order process. Based on the reaction coordinate derived from DFT calculations (Section 5, SI) and the dilution experiments that are brought below, complexation reaction proceeds through a low energy intermediate precomplex (HG)1:

H + G ⇌ (HG) ⇌ (HG)
This low-energy pathway is visualized in Figure S18 B. In order to find the association reaction rate order, we performed a series of dilution experiments ( Figure S19) near the coalescence temperature of NaPF6. The peak shapes and hence the rate of the reaction did not change on diluting the sample, which implies presaturation and the low energy precomplex formation according to Figure S18 B and therefore also a first-order association reaction.
A B Figure

WINDNMR fits
The 2ax proton signals for NaPF6 were chosen for simulations as those protons have good lineshapes over a large range of temperatures and are separated from other protons with at least one of the peaks uninterrupted by other peaks at all temperatures. For NaSbF6, the 3eq proton was observable over the whole temperature range, whereas 2ax proton was significantly broadened at higher temperatures. Figure S20. Simulated and experimental spectra of the variable temperature study as obtained from WinDNMR lineshape analysis for NaPF6. Figure S21. Simulated and experimental spectra of the variable temperature study as obtained from WinDNMR lineshape analysis for NaSbF6.   Figure S22. Eyring plot for the association process for SbF6 − (red circles) and PF6 − (black circles). Table S11. Temperature, sum of host-guest complex formation reaction rate constants, individual association reaction rate constants and natural logarithms used for the Eyring equation for 2ax proton for NaPF6. Coalescence region was not used for obtaining fits.

Computational details
Because the guest molecules are not chiral the calculated complexation energies do not depend on the chirality of the host, the geometry of (all-S)-cyclohexanohemicucurbit [8]uril was chosen from the previous computational work of Prigorchenko et al. 1 Geometry optimization for local minima were carried out using the BP86 14 functional, along with the def2-SV(P) 15 basis set. Energies were refined using the BP86-D functional, along with the def2-TZVPDref 15 basis set. The heavier atoms were described with the inclusion of the appropriate Stuttgart pseudopotential. 16 Vibrational analysis calculations were performed to ensure all chosen geometries were at local minima. The solvation model COSMO 17 (methanol ε = 32.613) was used for calculations. The total energies were calculated using the Gibbs free energy correction from the vibrational analysis. Gibbs free energy was estimated using the temperature 293.15 K and the pressure 0.1 MPa (if not stated otherwise). The calculations were performed using the program package Turbomole 6.5 18 . The maps of electrostatic potential (MEPs) were generated using Gaussian 09 19 . COSMO was not used for generating MEPs. The van der Waals radii for Sb (2.06), I (1.98) and Re (2.00) were added manually. 20 The results do not include a basis set superposition error correction due to the incompatibility between the continuum solvation model (COSMO) and the counterpoise approach to BSSE correction.

complex in MeOH
Number of solvent molecules in the cavity of cycHC [8] Encapsulated water molecules are higher in energy compared to the bulk water molecules. 21 This energy difference, in turn, affects the binding energy of the guest (high-energy solvent is released, guest-host complex forms, thus the total energy of the system is lower). The energy difference between the encapsulated and free solvent is governed by the cavity size (larger cavities allow more stable H-bonded networks) and the amount of molecules in it. Thus the number of the solvent molecules inside the cavity and the structure of the encapsulated solvent molecules play an important role in studying the formation of a host-guest complex.
According to the crystal structure of cycHC [8], the number of solvent molecules (methanol) captured in the cavity of cycHC [8] varies from one to three. 1 Our previous research 22 indicates that the nitrogen atoms are the strongest hydrogen bond acceptors of the cycHC [8], thus the captured methanol forms a hydrogen bond with cycHC [8] through its hydrogen atom as depicted on Figure S23 a. A search for binding positions of second methanol was conducted and eight positions of two methanol molecules were considered as depicted on Figure S24. Figure S24. Potential solvent positions within the cavity of cycHC [8]. Methanol molecules in the cavity of cycHC [8] are blue and red. The top row (molecules in parentheses) is the cycHC [8] viewed from side. The bottom row (molecules inside a circle) is the cycHC [8] viewed from top. Dashed lines represent a hydrogen bond between the methanol and cycHC [8].
The geometry where the second methanol molecule would form a hydrogen bond with the first methanol molecule ( Figure S24 d) was energetically the most favoured geometry (and most calculations converged into d).
The energies can be found in Table S12. The optimized geometry of structure d is depicted on Figure S25. As can be seen from the results (Figure S25 b), the hydrogen bond chain is preferred, thus the third methanol molecule would be connected via a hydrogen bond to the second one as depicted on Figure S26. 23 Figure S26. (a) (3MeOH)@cycHC [8] complex, where cycHC [8] is visualized without hydrogen atoms and hydrogen bonds as black lines; (b) (3MeOH)@cycHC [8] complex visualized with van der Waals viewed form the both portals. The red circle shows a cavity for a potential fourth MeOH molecule.
Although no fourth solvent molecule was observed in the experiment, the existence of a fourth solvent molecule was tested since the cavity potentially had room for a fourth solvent molecule (Figure S26 b).
Two cases were considered: the fourth solvent molecule is going to form a hydrogen bond with the third solvent molecule; the four solvent molecules form a hydrogen bond chain through the cavity as depicted on Figure S27. The second option was energetically favoured (Table S13).  Figure S27. (a) the cross-section of (4MeOH)@cycHC [8] complex where four solvent molecules form a hydrogen bond chain through the cavity; (b) the cross-section view of (4MeOH)@cycHC [8] complex where the fourth solvent molecule forms a hydrogen bond with the third solvent molecule.
The total energies of the optimized geometries on different temperatures were compared to evaluate the number of solvent molecules in the cavity of cycHC [8]. The total energy was composed of the SCF energy and the Gibbs free energy correction (or zero point energy correction, where necessary). The total energies were compared for: (1MeOH)@cycHC [ (Table S14). The results are depicted on Figure S28, where all energies are compared to the geometry of (1MeOH)@cycHC [8] on 300 K. At the room temperature the geometry with one solvent molecule was preferred. Thus, the complex with one methanol molecule ( Figure S23) was chosen for further studies.  Solvent molecule ejection and PF6 -encapsulation to cycHC [8] To study the guest exchange reaction (the solvent molecule ejection and the PF6 -ion encapsulation) a series of geometry optimization calculations were done with partially frozen geometry (BP86/def2-SV(P)). The trajectory of the PF6 -ion is depicted on Figure S29, where the frozen atoms are marked with an asterisk (*). All other atoms, including fluorine atoms were free to move during the optimization. The energy during the trajectory of the PF6 -anion was calculated starting from the distance of 12.0 Å from the centre of the cavity and with each next calculation, the ion was brought closer to the centre of the cavity by 0.2 Å. The found minima were further studied using the BP86-D dispersion corrected functional.

Optimized minima of the reaction
The geometries a, c and e on Figure S30 were taken for further study. They were optimized on BP86/SV(P) level of theory and vibrational analysis was performed to ensure all chosen geometries were at local minima. The energies were refined on BP86-D/TZVPD level of theory and the reaction path (without transition states) is depicted Figure  The geometry of o-a is depicted on Figure S32. During the optimization of o-a, the anion distance from the cavity center remained approximately 12.0 Å. According to Figure S30, the anion should not interact with the host. The methanol inside the cavity is sharing a hydrogen bond with the host. The geometry of o-c is depicted on Figure S33. The anion is situated on the opening of cycHC [8] (3.9 Å from the center of the cavity), while the methanol molecule is in the same position as in o-a. The geometry of o-e is depicted on Figure S34. After the hydrogen bond between the methanol molecule and the host is broken, the anion moves into the center of the cavity and replaces the methanol molecule. The methanol molecule is pushed out of the cavity, but it forms a new hydrogen bond with the PF6 − ion and stays on the opening of cycHC [8]. The anion adopts the same position as in the crystal structure. Figure S32. Minima of PF6 -ion closing in on cycHC [8].