Formation and Trapping of the Thermodynamically Unfavoured

Amplification of a thermodynamically unfavoured macrocyclic product through the directed shift of the equilibrium between dynamic covalent chemistry library members is difficult to achieve. We show for the first time that during condensation of formaldehyde and cis-N,N'-cyclohexa-1,2-diylurea formation of inverted-cis-cyclohexanohemicucurbit[6]uril (i-cis-cycHC[6]) can be induced at the expense of thermodynamically favoured cis-cyclohexanohemicucurbit[6]uril (cis-cycHC[6]). The formation of i-cis-cycHC[6] is enhanced in low concentration of the templating chloride anion and suppressed in excess of this template. We found that reaction selectivity is governed by the solution-based template-aided dynamic combinatorial chemistry and continuous removal of the formed cycHC[6] macrocycles from the equilibrating solution by precipitation. Notably, the i-cis-cycHC[6] was isolated with 33% yield. Different binding affinities of three diastereomeric i-cis-, cis-cycHC[6] and computational and Abstract Amplification of a thermodynamically unfavoured macrocyclic product through the directed shift of the equilibrium between dynamic covalent chemistry library members is difficult to achieve. We show for the first time that during condensation of formaldehyde and cis - N,N  -cyclohexa-1,2-diylurea formation of inverted-cis -cyclohexanohemicucurbit[6]uril ( i-cis -cycHC[6]) can be induced at the expense of thermodynamically favoured cis -cyclohexanohemicucurbit[6]uril ( cis -cycHC[6]). The formation of i-cis- cycHC[6] is enhanced in low concentration of the templating chloride anion and suppressed in excess of this template. We found that reaction selectivity is governed by the solution-based template-aided dynamic combinatorial chemistry and continuous removal of the formed cycHC[6] macrocycles from the equilibrating solution by precipitation. Notably, the i-cis -cycHC[6] was isolated with 33% yield. Different binding affinities of three diastereomeric i-cis -, cis -cycHC[6] and their chiral isomer ( R,R )-cycHC[6] for trifluoroacetic acid demonstrate the influence of macrocycle geometry on complex formation. diastereoselectivity of the formation of 2 and towards 2 by increasing the concentration of HCl in the reaction mixture. The combined experimental and DFT study of the formation of macrocycles revealed that the ratio of diastereomeric cycHCs 2 and 1 depends on an interplay of three important steps during the synthesis: first, the thermodynamic control over templated oligomerisation and macrocyclisation, second, kinetically favoured and non-selective macrocyclisation step and third, the solubility equilibrium of cycHCs. A correlation between the geometry of diastereomeric cycHCs and affinity towards TFA was demonstrated. This study contributes to an understanding of the formation mechanism and binding capacity of cycHC[6]s that can be utilized in supramolecular and catalytic applications of single-bridged cucurbiturils.

Amplification of a thermodynamically unfavoured macrocyclic product through the directed shift of the equilibrium between dynamic covalent chemistry library members is difficult to achieve. We show for the first time that during condensation of formaldehyde and cis-N,N'-cyclohexa-1,2-diylurea formation of inverted-cis-cyclohexanohemicucurbit [6]uril (i-cis-cycHC [6]) can be induced at the expense of thermodynamically favoured cis-cyclohexanohemicucurbit [6]uril (cis-cycHC [6]). The formation of i-cis-cycHC [6] is enhanced in low concentration of the templating chloride anion and suppressed in excess of this template. We found that reaction selectivity is governed by the solution-based template-aided dynamic combinatorial chemistry and continuous removal of the formed cycHC [6] macrocycles from the equilibrating solution by precipitation. Notably, the i-cis-cycHC [6] was isolated with 33% yield. Different binding affinities of three diastereomeric i-cis-, cis-cycHC [6] and their chiral isomer (R,R)-cycHC [6] for trifluoroacetic acid demonstrate the influence of macrocycle geometry on complex formation.

Introduction
Reactions forming multiple chemical bonds simultaneously are challenging. Efforts to develop such reactions can, however, be rewarded by leading to complex target compounds in a single step. Many examples of single-step synthesis of oligomeric macrocycles have been described, yet achieving product selectivity often remains an obstacle [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15] . Templates can offer control over product formation through preorganization of a corresponding linear precursor, or through altered thermodynamic equilibrium of the reaction. For example, we recently demonstrated that macrocyclic chiral hemicucurbiturils (HC) can selectively be obtained both in solution 16,17 and in the solid state 18 , with the product determined by the choice of the anionic template in the reaction media. The parent compounds, double-bridged cucurbiturils (CBs), 7,[19][20][21][22] are also synthesized in a single step, but the product distribution is not as efficiently influenced by the addition of templates 23 . Currently, four-24-29 , six-, eight-30 and twelve 16 -membered HCs are known, with the largest variation in substituents achieved for the six-membered 16,24-29,31-35 HCs. The cavity shape and electronic structure of HCs vary, leading to distinct applications in host-guest systems [36][37][38][39][40][41][42][43][44][45][46][47][48] . To date, no reports on the formation of stereoisomeric HCs from the same monomer units have been presented. Herein, we report for the first time the amplification of formation of a thermodynamically unfavoured inverted-cis-cyclohexanohemicucurbit [6]uril 1 (Figure 1) through dynamic covalent chemistry. Such a single monomer modification and formation of diastereomeric macrocycles lead to the enrichment of the pool of hosts, which is important for the development of highly selective host-guests interactions. Thorough NMR and single-crystal XRD analysis revealed that the new cycHC [6] had one monomer in an inverted configuration. Consequently, the new macrocycle was named inverted-cis-cyclohexanohemicucurbit [6]uril 1. Similarly to 2, its diastereoisomer 1 is achiral, however, the latter contains only a single symmetry element-a plane of symmetry ( Figure 2). Therefore, compared with the highly symmetric HC 2, in which the 13 C NMR signals of all six monomers overlap (Figure 2a), the carbon spectrum of 1 reflects the differences of the urea monomers in the macrocycle (Figure 2b). The uneven distribution of electron density in 1 is demonstrated by large variations in the 13 C-NMR chemical shifts. Single crystals of compound 1 (Figure 1) were obtained from deuterated chloroform. The location of the inverted monomer appears as disordered between four positions in hexameric 1. Two disorder components (sum of 0.5) corresponding to the inverted monomer were located in the asymmetric unit, which consists of half of the 1 molecule. The remaining two components are generated by symmetry. Inversion of the monomer does not affect the direction of hydrogen bonding between i-cis-cycHC [6] 1 and the solvent (SI, Figure S8 B), nor does it significantly change the shape of the molecule, which results in the extensive disorder observed in the crystal structure and apparent isostructurality to cis-cycHC [6] 2. The first inverted HC 1 can be obtained in up to 33% isolated yield (Table 1). This contrasts the synthesis of analogous inverted double bridged CBs, the i-CB [6] and i-CB [7], which were isolated in 2.0% and 0.4% yield, respectively. 51 Formation of diastereomeric macrocycles from achiral cis-(R,S)-monomers proceeds through desymmetrisation of the monomer, resulting in either R,S,S,R-or R,S,R,S-ordered oligomers 4 and 5 (Scheme 1). Such desymmetrisation can in principle occur with all HCs, where the monomers are not C2-symmetric. Trapping of an inverted isomer hence suggests also novel opportunities for the synthesis of non-centrosymmetric HCs, from achiral monomers.
In the described synthesis conditions, Wu et al. 30 isolated 2 in 78% yield, however in our hands the same conditions led to the formation of 2 and 1 in nearly equal amounts. The macrocycles were isolated in an overall 55-77% yield, with the ratio of 2 to 1 varying from batch-to-batch in the range of 1.4:1 to 1:1.4 (Table 1, line 1). We have previously shown that the formation of chiral cycHCs occurs through the dynamic covalent chemistry and macrocyclic products are thermodynamically stabilized by suitably shaped and sized anionic templates. 18,30 The potential ability of isomer 1 or 2 to rearrange into a more thermodynamically stable diastereomer was checked by subjecting both 2 and 1 independently to analogous reaction conditions (Table 1, lines 5 and 6). A very small portion of 2 was observed to react in 4 M HCl, whereas nearly two-thirds of 1 was converted to the more symmetric 2. This clearly indicated that cis-cycHC [6] 2 is the thermodynamically more stable diastereomer of the two, with i-cis-cycHC [6] 1 undergoing reversible conversion in the presence of aqueous hydrochloric acid. The analogous i-CBs were shown to be kinetic intermediates, that can be converted to thermodynamically more stable CBs. 51 However, increasing the temperature, from 70 o C to 80 o C under microwave (MW) or to 110 o C in an oil-bath, which should accelerate the reaction and lead to faster accumulation of the thermodynamically more stable product 2, appeared not to affect the ratio of the formed cycHC diastereomers (Table 1, lines 2 and 3). This ratio remained constant even if the reaction time was increased from 4 to 17 hours (see SI, Table S1). The amount of 2 was, however, clearly increased by increasing the concentration of HCl from 4 M to 8 M (Table 1, line 4), indicating that 2 is preferably formed in higher template Cl − and acid concentrations. To gain a better understanding of the stability of the involved oligomers and macrocycles, that lead to the trapping of 1, computational studies were performed. A DFT study at the BP86/def-TZVP level 52-55 (see SI) of model structures of R,S,S,R-and R,S,R,S-ordered dimers 6 and 7 (Scheme 1; SI, Figure S9), showed 7 to be 3 kJ/mol higher in energy. The gap between energies of the C-shaped and S-shaped dimers 6 and 7, respectively, was even further increased to 14 kJ/mol by complexation with the chloride anion ( Figure 3). To distinguish between geometric and electronic influences, the chloride was deleted from the optimized geometry and a single-point calculation was performed. Without the chloride, the unoptimized S-shaped dimer was 6 kJ/mol lower than the C-shaped one. This confirms that the C-shaped form is stabilized by the interactions with the chloride ion. These results reflect that the chloride anion is templating the formation of C-shaped oligomers, and therefore in 2 and 1 the R,S,S,Rordered configuration prevails. Higher stability of C-shaped oligomers is similar to findings from core cucurbituril C-and S-shaped dimers 56 which showed that C-shaped dimers are thermodynamically favoured. A computational study on the diastereomeric cycHC [6] macrocycles indicated that compound 1 has 17 kJ/mol higher energy than 2 and upon encapsulation of the chloride anion, the energy difference increased to 30 kJ/mol (see the structure of Cl -@1 in Figure  4A). This demonstrates that the formation of the inverted diastereomer is thermodynamically unfavourable and its formation is especially hampered if the Cl¯@2 complex is formed. This result is consistent with the experimental outcome, as the proportion of 2 depends principally on the concentration of hydrochloric acid; however, it does not explain the equimolar formation of the unfavourable inverted isomer 1 in 4 M HCl.
Modelling of the formation of 1 and 2 from six-membered iminium intermediates unexpectedly showed, that all the studied hexameric iminium chloride geometries formed HCl@cycHC [6] inclusion complexes upon energy minimization (see SI for details). Therefore, we can state that the final macrocyclisation step proceeds without a transition state. Even more significantly, the resulting inclusion complexes of the diastereomeric macrocycles with hydrochloric acid were energetically very similar. The lowestenergy geometry of the HCl inclusion complex of 1, which is depicted in Figure 4B, was even 1.5 kJ/mol lower than that of 2. This outcome indicates that in the final macrocyclisation step, the formation of inverted diastereomer 1 is as favourable as the formation of 2. The fact that the reactions end in a heterogeneous mixture suggests that the kinetic product 1 is trapped by precipitation. Low solubility of the macrocyclic products 1 and 2 does not allow for the equilibration of 1 to the thermodynamically more favourable macrocycle 2.. To probe the influence of geometry of cycHCs on their application as hosts, binding of the three diastereomers 1, 2 and 8 and their methylated monomers the (R,S)-and racemic (R*,R*)-(N,N-dimethyl)-cyclohexa-1,2-diylureas, 9 and 10 to trifluoroacetic acid (TFA) was compared ( Table 2). TFA was selected as a guest because it serves as an efficient template for conversion of (R,R)-cycHC [6] 8 to (R,R)-cycHC [8] 30 . The stoichiometry of the binding of urea derivatives with TFA was assessed by Job's method, which indicated a 1:2 stoichiometry for cycHCs and 1:1 for mono-ureas 9 and 10 (See SI). Fitting of the cycHCs titration data to a twostep binding isotherm revealed that the second TFA molecule complexes with all macrocycles with positive cooperativity with interaction parameter 57 in the range of 3 to 9.  The affinity of the chiral cycHC 8 towards the first TFA molecule (K11, Table 2, line 3) was significantly lower than with 2 and 1. The second TFA and the cumulative association was strongest with 2, which has the largest distance between cyclohexane rings at the portals. (Table 2, line 2, Figure 1). These observations reflect the correlation between the geometry and size of the openings of cycHC [6]s and acid binding ( Table 2, lines 1 to 3, Figure 1). Following the protons that point inside and outside of the cavity of macrocycle (see SI, Figure S13), confirms that external complexes with fast exchange on the NMR time scale are formed. Previously, 1:1 and 1:2 binding of carboxylic acids and carboxylates have also been studied for other single-bridged cucurbiturils, the bambus [6]urils 40,58,59 .

Conclusions
Our research shows that the first inverted HC 1 can be isolated with up to 33% yield and the diastereoselectivity of the formation of 2 and 1 can be shifted towards 2 by increasing the concentration of HCl in the reaction mixture. The combined experimental and DFT study of the formation of macrocycles revealed that the ratio of diastereomeric cycHCs 2 and 1 depends on an interplay of three important steps during the synthesis: first, the thermodynamic control over templated oligomerisation and macrocyclisation, second, kinetically favoured and non-selective macrocyclisation step and third, the solubility equilibrium of cycHCs. A correlation between the geometry of diastereomeric cycHCs and affinity towards TFA was demonstrated. This study contributes to an understanding of the formation mechanism and binding capacity of cycHC [6]s that can be utilized in supramolecular and catalytic applications of single-bridged cucurbiturils.    Unless otherwise stated, all reagents were purchased from commercial suppliers and used as received. Solvents used for flash chromatography were reagent grade, which were dried and distilled prior to use according to standard procedures. (R,S)-N,N'-cyclohexa-1,2-diylurea 3 was synthesized starting from either cis-1,2-diaminocylohexane or mixture of cis-and trans-1,2diaminocylohexane according to literature procedure 1 . Reactions under microwave were performed in CEM Discover ® microwave reactor. Flash chromatography was run over Thomar CC Silica Gel 60 (0.04-0.063 mm) stationary phase. Infrared spectra were obtained on a Bruker Tensor 27 FT-IR spectrometer and are reported in wavenumbers. HPLC based reaction rate monitoring was performed on an Agilent 1200 Series HPLC system with a Kinetex C18 column (2.1x100 mm, 2.6 µm) and UV-detection at 210 nm. Identification of reaction products and traces was performed by RP-HPLC-HRMS on an Agilent 6540 UHD Accurate-Mass Q-TOF LC/MS spectrometer with a Zorbax Eclipse Plus C18 column (2.1x150 mm, 1.8 µm) and AJ-ESI ionization. 1D 1 H and 13 C-NMR spectra were acquired on a Bruker AvanceIII 400 MHz spectrometer or a Bruker Avance III 800 MHz. Chemical shifts were referenced to the chloroform residual solvent signal in 13 C 77,160 ppm and 1 H 7,260 ppm.

Notes and references
General synthetic procedure for 1 and 2 from (R,S)-N,N'-cyclohexa-1,2-diylurea: Suspension of (R,S)-N,N'-cyclohexa-1,2-diylurea 3 (prepared according to procedure in ref. 1) and paraformaldehyde was heated in aqueous HCl acid on oil bath. All reaction mixtures were heterogeneous, ratio of formed macrocycles 2 and 1 was determined from crude reaction mixture by HPLC-UV analysis (See SI section "Quantitative analysis of cycHCs"). Reaction mixture was filtered, washed with water and dried in open air. Crude product was dissolved from the filter with CH2Cl2, solvent evaporated and crude product was purified and macrocycles 1 and 2 were separated by flash chromatography on silica gel, gradient elution with 1-5% of i-PrOH in CH2Cl2.  Figure S1. HPLC-UV chromatogram of crude reaction mixture performed in A) 4 M and B) 8 M HCl (see Table S1 row 5 and 8, and sample preparation details in next section).

S4
The formation of cis-cycHC [8] starting from cis-cycHC [6] 2 was also investigated. cycHC 2 was subjected to following reaction conditions: mixture of formic acid and acetonitrile, mixture of acetic acid, acetonitrile and NaPF6 and mixture of trifluoroacetic acid and acetonitrile, all at room temperature and analogously to our previous study for formation of (R,R)-cycHC [8] 4 . MS analysis of crude reaction mixtures confirmed formation of other homologues form cis-cycHC [6] (See Figure S2), but these compounds were formed in very low yield and therefore their isolation was not attempted.  For chromatographic analysis three samples (20 µl) of the reaction mixture as a suspension were taken for three parallels and diluted in CHCl3/CH3OH (1:4) mixture to reach fully dissolved sample. The concentrations were found using the equation y=ax+b equation from the calibration curve (Table S2).
Synthetic procedure of inter-conversion of cycHCs
Inverted-cis-cycHC [6] 1 has one plane of symmetry, which simplifies the analysis of its NMR spectra. As two halves of the molecule are equivalent, we will consider only 4 monomers (, γ, δ), where denotes the inverted monomer. It is also important that monomers are in a zig-zag orientation and therefore numeration of the carbon atoms of the subsequent monomers alternates in each direction. Figure S4. Part of a COSY spectrum, indicating neighbouring hydrogens, marked by arrows on the structure. The respective crosspeaks are indicated by circles in the figure.

S10
The COSY spectrum in Fig. S4 indicates that 1 and 5 as well as 1 and 5 hydrogen atoms show a correlation cross peak and are therefore hydrogen atoms of the same monomer. These pairs of hydrogen atoms are marked with arrows and the respective cross peaks are circled in Fig. S4. The 1 and 1 hydrogen atoms originate from symmetrical monomers and give only a single peak in the 1 H NMR spectrum and therefore no cross peaks in the COSY spectrum. Figure S5. Parts of a HSQC and a HMBC spectra. The correlations in the spectrum are marked by arrows on the molecular structure.
The HSQC signals in Fig. S5 indicate correlations between 10, 10 and 10 bridge carbon atoms and the respective hydrogen atoms. Moreover, the HMBC spectrum in Fig. S5 shows that 1 and 1 hydrogen atoms both give a correlation with 10 bridge carbon atom and therefore must be next to the same bridge moiety in the macrocycle. From the earlier analysis of the COSY spectrum we know that the 1 hydrogen atoms belong to a symmetrical monomer. We have denoted 1 and 1 hydrogens and their correlation with 10 carbon atoms with blue arrows in Fig.  S5. The same logic applies to 1 and 1 hydrogen atoms with these hydrogen atoms giving a S11 correlation with 10 bridge carbon atom, which is marked with black arrows; here 1 hydrogen atoms are from the symmetrical monomer. Furthermore, 5 and 5 hydrogen atoms give a correlation in the HMBC spectrum with 10 bridge carbon atoms, which can be marked with green arrows in Fig. S5.
The NOESY spectrum in Fig. S6 indicates cross peaks between 1 and 1hydrogen atoms as well as between 1 and 5 hydrogen atoms, showing that all these hydrogen atoms point towards the cavity of the macrocycle. There is a cross peak between 5 and 5 hydrogen atoms, demonstrating the same logic. These correlations map well with the positions of the hydrogen atoms in the crystal structure, also demonstrated in Fig. S6. No cross peaks are observed for 1 hydrogen atoms, which confirms that these hydrogen atoms point away from the cavity of the macrocycle.  The rest of the hydrogen and carbon atoms were assigned in analyzing the COSY and the HSQC spectra and their chemical shifts and coupling constants are brought further.  Figure S7. 13 C-NMR spectra of i-cis-cycHC [6] 1 with atom numbers. S13 1 H and 13 C-NMR spectra of compounds 1, 2, 9, 10. S16 1 H and 13 C-NMR spectra of rac. (R*,R*)-(N,N'-dimethyl)-cyclohexa-1,2-diylurea 10 S17

Crystallographic analysis of inverted-cis-cycHC[6] 1
Single crystal X-ray diffraction data was collected at 123K on Rigaku Compact HomeLab diffractometer, equipped with a Saturn 944 HG CCD detector and Oxford Cryostream cooling system using monochromatic Cu-Kα radiation (1.54178Å) from a MicroMax TM -003 sealed tube microfocus X-ray source. The strategy of data collection was calculated using RigakuCollectionStrategy 7 . CrysAlisPro 8 was used for data reduction and empirical absorption correctionusing spherical harmonics implemented in SCALE3ABSPACK scaling algorithm 9 . The structure was solved using SHELXT 10 and refined by full-matrix least-squares method against F 2 with SHELXL-2014 10 through OLEX2 11 program package. All non-hydrogen and nondeuterium atoms were refined with anisotropic atomic displacement parameters. Hydrogen atoms attached to carbon atoms were treated as riding atoms, using isotropic displacement parameters Uiso(H) = 1.2Uiso(C) for CH and CH2. The figures were drawn using the program Mercury CSD 3.9 12 and POV-Ray 3.7 13 .

Crystal data
C54H72Cl18D6N12O6 The compound crystallizes in a centrosymmetric space group P-1, with half of the macrocycle and three solvent molecules in the asymmetric unit. An inversion center is located at the centre of the cavity of i-cis-cycHC [6] 1. The position of the inverted monomer in the hexamericcycHC [6] appears to be disordered between four locations (Figure S8 A). Two independent disorder components were found for two neighboring monomers in the asymmetric unit. The relative occupancies of these were refined freely with the sum of site occupation factors restrained to be a constant at 0.500, resulting in sof 0.330(2) and 0.170(2) for the two sites respectively. The remaining two disorder components are generated by symmetry around the inversion center.
The packing of the macrocycles seems not to be affected by the position of the inverted monomer, as the crystal structure is largely isostructural to the cis-cycHC [6] previously reported in the literature (CCDC 716122). This is most likely due to the fact that the orientation of the hydrogen bond accepting carbonyl group in the inverted monomer is not drastically changed compared to the other monomers, so the hydrogen bonding to the solvent is not altered by the presence of the inverted monomer (Figure S8 B). S18 Figure S8. . The relatively large anisotropic displacement ellipsoids of these solvent molecules and the remaining electron density (max 0.47) in close vicinity indicate further disorder, but modeling more than three disorder components per site was not attempted. The third chloroform molecule in the asymmetric unit was modeled in two disorder components, with the relative occupancy of the two components refined freely, giving sof 0.788(9)/0.212 (9).

S19
Computational studies

General information
The effect of chloride ion on different stages of the formation of cyclohexanohemicucurbit [6]urils was studied computationally. The input coordinates of cis-cycHC [6] 2 and i-cis-cycHC [6] 1 were generated from the X-ray crystal structures of the molecules. All geometries were fully optimized, and stationary points characterized with vibrational analysis. In the case of the system HCl@1 B (described below), a residual imaginary vibration of 3.03 cm -1 remained; all other systems were characterized as true minima. Vibrational zero-point energies were added to electronic energies and the resulting differences evaluated. All the calculations were performed with the density functional theory, using the functional BP86, the basis set def2-TZVP and the program package Turbomole 6.3. 14

Methylated dimers
From the optimized isomeric cyclic structures both R,S,S,R-and R,S,R,S-ordered dimers were isolated and their methylated geometries (6 and 7) optimized ( Figure S9). Next, the effect of the presence of chloride ion close to named dimers was modelled, adding the chloride ion "inside" of the dimer, where it would mimic the chloride ion's position being in the cavity of the cycles. From the optimized Cl ‾ @R,S,S,R-ordered dimer and Cl ‾ @R,S,R,S-ordered dimer complexes the chloride ions were then removed and single-point calculations (SPC) for both dimers were performed. Figure S9. DFT models of dimers 6 (C-shaped and R,S,S,R-ordered) and 7 (S-shaped and R,S,R,S-ordered).

Iminium intermediates
Initial geometries for seven different iminium intermediates were obtained from optimized geometries of corresponding closed cycles ( Figure S10a) via moving atoms apart at the appropriate C-N bond (shown on Figure S10b). The bond was broken by changing a bond angle at the opposite side of the macrocycle and adding a proton to the nitrogen atom of the broken bond. Then the chloride ion was positioned in the middle of the cavity of all different iminium intermediates ( Figure S10b) and the reactions between the intermediates and the chloride ion were modelled. In the structure of cis-cycHC [6] 2 all C-N bonds in methylene bridges are equivalent, therefore only a single iminium structure with encapsulated Cl‾ anion was built. The inverted isomer has six non-equivalent C-N bonds, therefore six iminium chloride complexes were constructed. All the studied oligomeric iminium chloride geometries formed HCl@cycHC [6] inclusion complexes upon energy minimization. All the HCl@i-cis-cycHC [6] S20 complexes formed from the "pairs" of the iminium intermediates of i-cis-cycHC [6] 1 (A1 and A2, B1 and B2, C1 and C2) had the same final structures with almost equivalent energies. Therefore as a result of seven different cyclisation reactions only four different HCl@cycHC [6] complexes were formed: one HCl@cis-cycHC [6] and three HCl@i-cis-cycHC [6] ( Figure S10c). The final structures and energies are given for the complexes with the lowest energies, as the energy difference between the same complexes formed from different input geometries (for example complex A formed from either A1 or A2) was below 1 kJ•mol -1 . Figure S10. a) A schematic representation of cis-cycHC [6] 2 and i-cis-cycHC [6] 1 with inverted monomer depicted with outward-projecting hydrogens. b) A schematic representation of the formation of seven iminium intermediates, where the dotted red line indicates the location of the broken C-N bond. After the breakage of the bond a proton was added to the nitrogen atom of the broken bond and chloride ion into the cavity of the intermediate. c) HCl@cis-cycHC [6] and three HCl@i-cis-cycHC [6] complexes showing the position of HCl in formed complexes. All the pairs of the intermediates (A1 and A2, B1 and B2, C1 and C2), when supplemented with a chloride ion, led to formation of the same HCl@i-cis-cycHC [6] complex, A, B, or C, respectively.  Figure S12. Job plot for 5 mM solutions of 1 with TFA, following protons of 1 protons at 3,1 ppm (δ1) and 4,6 ppm (α10) by 1 H-NMR (see S9-S13 for assignment) Figure S13. Job plot for 5 mM solutions of monourea 9 with TFA, following TFA by 19 F-NMR.

Determination of the association constants
General remarks 19 F-NMR titration experiments were conducted in deuterated chloroform at 297 K. The concentration of TFA solution in the beginning of titration was in between 0.97 to 1.75 mM and urea derivatives 1, 2, 8, 9 and 10 stock solution in between 70 to 33 mM. Urea was added by syringe in small portions and the spectra recorded after each addition, resulting in a set of spectra (15-18 per titration). The 19 F-NMR spectra were recorded on a Bruker Avance II 400MHz spectrometer using regular 5 mm NMR tubes. The progressive changes of the chemical shifts of TFA were followed. In order to avoid interactions with chemical shift reference compound, 19 F TFA chemical Binding of TFA to cycHC [6]s can be expressed as follows: TFA + cycHC [6] K11 → TFA • cycHC [6]  Binding of TFA to monoureas 9 and 10 can be expressed as follows: TFA + urea derivative K11 → TFA • urea derivative Association constants K11 and K12 were calculated using free online tools from http://supramolecular.org/ (accessed March 22, 2018).  10+TFA K1=21•10 2 error 8.1% S40 1 H-NMR titration of (R,R)-cycHC [6] 8 with TFA Figure S15. 1 H-NMR of titration of (R,R)-cycHC [6] 8 with TFA. Identification of 1 H-NMR signals is reported previously 15 . Proton 1ax at 2.77 ppm shifts upon addition of 16 eq. of TFA to 2.86 ppm. Therefore external binding is proposed as proton 1ax. which is positioned outside of the cavity. chemical shift value changes and 2ax at 2.40 ppm. which is positioned inside the cavity. does not change upon addition of TFA.