Orthogonal, modular anion–cation and cation–anion self-assembly using pre-programmed anion binding sites

Subcomponent self-assembly relies on cation coordination whereas the roles of anions often only emerge during the assembly process. When sites for anions are instead pre-programmed, they have the potential to be used as orthogonal elements to build up structure in a predictable and modular way. We explore this idea by combining cation (M+) and anion (X−) binding sites together and show the orthogonal and modular build up of structure in a multi-ion assembly. Cation binding is based on a ligand (L) made by subcomponent metal-imine chemistry (M+ = Cu+, Au+) while the site for anion binding (X− = BF4−, ClO4−) derives from the inner cavity of cyanostar (CS) macrocycles. The two sites are connected by imine condensation between a pyridyl-aldehyde and an aniline-modified cyanostar. The target assembly [LM-CS-X-CS-ML],+ generates two terminal metal complexation sites (LM and ML) with one central anion-bridging site (X) defined by cyanostar dimerization. We showcase modular assembly by isolating intermediates when the primary structure-directing ions are paired with weakly coordinating counter ions. Cation-directed (Cu+) or anion-bridged (BF4−) intermediates can be isolated along either cation–anion or anion–cation pathways. Different products can also be prepared in a modular way using Au+ and ClO4−. This is also the first use of gold(i) in subcomponent self-assembly. Pre-programmed cation and anion binding sites combine with judicious selection of spectator ions to provide modular noncovalent syntheses of multi-component architectures.


General Comments
All reagents were obtained from commercial suppliers and used as received unless otherwise noted. The aminocyanostar macrocycle (NH2-CS) was synthesized from iodo-cyanostar (CS-I) according to the reported literature procedure. 1 Column chromatography was performed on silica gel (40-63 μm). Thin layer chromatography (TLC) was performed on pre-coated silica gel plates (250 μm thick, TLG-R10011B-323, Silicycle, Canada) and observed under UV light.
Trifluoromethyl toluene (Ph-CF3) was used as an internal standard for the collection of 19 F NMR spectra. The 19 F NMR peak of Ph-CF3 in CD2Cl2 was referenced at -62.93 ppm based on the CFCl3-calibration studies for 19 F NMR references in various deuterated solvents. 2 All 19 F NMR spectra were collected below 10 mM to eliminate any concentration effects. 2 For all 1 H NMR titrations a solution of the cyanostar derivative was placed in an NMR tube, sealed with a rubber septum and an initial 1 H NMR spectrum was recorded. Aliquots of a solution of the corresponding salt were then added with a microsyringe through the rubber septum. 1 H NMR spectra were recorded after each addition.
For 19 F NMR titrations, a CD2Cl2 solution of the cyanostar derivative was placed in an NMR tube, 10 µL of trfluorotoluene (Ph-CF3) was added in the solution for standardization of chemical shifts of 19 F peaks. Afterwards, the NMR tube was sealed with a rubber septum and an initial 19 F NMR spectrum was recorded. Another 1 H NMR spectrum was also recorded to verify that the Ph-CF3 had no interaction with the cyanostar derivative. Aliquots of a solution of the corresponding salt were added with a microsyringe through the rubber septum and 19 F NMR spectra were recorded after each addition.
Nuclear magnetic resonance (NMR) spectra were recorded on a Varian Inova (600 MHz, 500 MHz, 400 MHz) and Varian VXR (400 MHz) spectrometers at room temperature (298 K). Chemical shifts were referenced to residual solvent peaks. The pulse gradient spin echo (PGSE) experiments provided the diffusion coefficients. The NMR peaks were analyzed to determine diffusion coefficients using Vnmrj's analysis software. Average diffusion coefficients and errors were generated from multiple 1 H NMR peaks. High-resolution electrospray ionization (ESI) mass spectrometry was performed on a Thermo Electron Corporation MAT 95XP-Trap mass spectrometer.

Self-assembly of [POPCu-CS-BF4-CS-CuPOP] + by Cation-Anion Pathway
In a 20-dram glass vial, [POPCu-CS]•BAr F 4 (6.2 mg, 0.0042 mmol, 1 equivalent) and TBABF4 (0.4 mg, 0.0013 mmol, 0.5 equivalent) were taken and 1 mL dichloromethane was added to the vial. A red solution was obtained. The solvents were evaporated by blowing argon to obtain [POPCu-CS-BF4-CS-CuPOP]•BAr F 4 as a red solid, which was then dissolved in CD2Cl2 and subjected to characterization. NMR spectroscopy showed the quantitative formation of [POPCu-CS-BF4-CS-CuPOP] + when compared to the 1 H NMR of the assembly obtained in one-pot ( Figure  S1, S2). The integration of 1 H NMR peaks are normalized to one cyanostar in assembly. 1

Self-assembly of [CS-BF4-CS] -
NH2-CS (3.5 mg, 0.004 mmol, 1 equivalent), F-PyCHO (0.92 mg, 0.008 mmol, 2 equivalent) and TBABF4 (3.9 mg, 0.012 mmol, 3 equivalent) were taken in a 20-dram glass vial and 1.0 mL dichloromethane was added in the vial. Excess TBABF4 was used in the reaction to ensure that all macrocycles form 2:1 cyanostar:BF4 − complexes in solution. A yellow solution was obtained. The solvents were evaporated by blowing argon. The solid precipitate was washed with Et2O (31 mL) and dried to obtain the final assembly (5.4 mg, 0.0022 mmol) under vacuum. The precipitate was dissolved in CD2Cl2 and the resulting solution was characterized by 1
We considered that the Au + ions may form a bridged complex based on similar work by Berry, Olmsted, and Balch, 10 but the high-resolution mass spectrometry ( Figure S35 ] + , has a theoretical m/z of 3120.1592, and a gold bridged assembly has a theoretical a m/z of 2858.0686. As a result, we likely have a mixture of assemblies in solution and cannot make any conclusive statements as to the nature of the complex formed by the anion-cation pathway only other than to say that it is different than the one-pot and cation-anion assemblies.

Self-assembly of [CS-BF4-CS]in Presence of Excess TBABF4
NH2-CS (3.5 mg, 0.004 mmol, 1 equivalent), F-PyCHO (0.46 mg, 0.004 mmol, 1 equivalent) and TBABF4 (6.6 mg, 0.020 mmol, 5 equivalent) were taken in a 20-dram glass vial and 1.0 mL dichloromethane was added in the vial. Excess TBABF4 was used in the reaction to ensure that all macrocycles form 2:1 cyanostar:BF4 − complexes in solution. A yellow solution was obtained. The solvents were evaporated by blowing argon, and the crude oil was dried under vacuum overnight. The crude product (8.6 mg) was directly dissolved in CD2Cl2 to take 1 H NMR of the assembly.

Self-assembly of Cyanostar-based Gold(I) Assembly by Anion-Cation Pathway in Presence of Excess TBABF4
In a 20-dram glass vial, TBA + [CS-BF4-CS] − (8.6 mg, 0.0004 mmol based on CS-NH2 moles, 1 equivalent), which had excess TBABF4 present, and [PPh3Au]•NTf2 (0.6 mg, 0.0008 mmol, 2 equivalents) were taken and 1 mL dichloromethane and 10 µL acetonitrile was added in the vial. A yellow solution was obtained. The solvents were then evaporated by blowing argon and the crude oily precipitate was washed with Et2O (2 mL × 3) and subjected to characterization.

Self-assembly of Gold(I) Model Complex
Para-toluidine (13.1 mg, 0.122 mmol, 1 equivalent), F-PyCHO (15.3 mg, 0.122 mmol, 1 equivalent) and [PPh3Au]•NTf2 (90.3 mg, 0.122 mmol, 1 equivalent) were dissolved in 12 mL dichloromethane in a 50 mL round-bottom flask and stirred overnight for 16 h at room temperature. Afterwards, all solvents were evaporated. The resulting yellow precipitate was dissolved in a minimum amount of dichloromethane, 20 mL Et2O was added in the solution, and kept in a freezer (−32 °C) until yellow single crystals of the model complex was obtained. The yellow crystals were separated and washed with Et2O (35 mL) to obtain the pure model complex (114 mg, 0.120 mmol, 98 %). Some of the crystals were used for the structural determination of the complex in the solid-state by single crystal X-ray diffraction studies. 1      Only intramolecular cross-peaks are observed in the dimer ( Figure S52). The chemical shifts of 1 He and 1 Hc are more similar in dimer ( Figure S52) than they are in the monomer complex ( Figure S50, S51), so no cross peaks are observed in dimer complex.

Diagnostic 1 H NMR Shifts upon Self-assembly
Imine formation and metal coordination show characteristic changes in the chemical shifts of key protons upon formation of the target assembly, [POPCu-CS-BF4-CS-CuPOP] + ( Figure S53). The aldehyde proton at 10.16 ppm ( Figure S53c) is consumed to form the imine with its proton (Hg) at the diagnostic position of 9.01 ppm ( Figure S53a). 12 Other proton resonances close to the metal binding site are characteristic of imine bond formation and copper complexation. 13 For example, the aniline's ring hydrogen (Hf, 6.84 ppm, Figure S53b) shifts downfield (7.23 ppm, Figure S53a) upon conversion to the imino phenylene, and the phosphine protons (Hk) shift modestly from 7.02 to 7.09 ppm ( Figures  S53a,d). The cluster of peaks at 8.76 ppm in NH2CS are assigned to the outer Hb protons of the macrocycle. After one-pot assembly, these shift upfield to 8.41 ppm matching the parent cyanostar. 14    Figure S56. Diagnostic 1 H NMR peaks (CD2Cl2, 0.5 mM) of gold(I) assemblies obtained by (a) anion-cation, (b) cation-anion, (c) one-pot self-assembly.

1 H NMR Titration Controls with Cyanostar
Parent cyanostar forms a 2:1 [pCS-BF4-pCS] − assembly. 14 A 1 H NMR titration monitoring the addition of TBABF4 to cyanostar ( Figure S50a, CD2Cl2) shows fast exchange peaks. Based on the Hb on cyanostar, we see conversion at 0.5 equivalents (Figure 50b). The inner cavity protons, Ha and Hd, shift downfield, and the outer cavity protons, Hb and Hc shift upfield (Figure 50a), which is consistent with BF4binding inside of the cyanostar cavity to form a 2:1 complex.
Ion pairing is known to play a role in CD2Cl2, and the inflection points of the Hα on TBA are consistent with this expectation ( Figure S50c). Between 0 and 0.5 equivalents, Hα on TBA + shifts upfield, which is consistent with the cyanostar binding BF4and disrupting ion pairing with the free TBA + . Upon the addition of excess TBABF4, Hα on TBA shifts downfield, which is consistent with the formation of ion pairs between TBA + and excess BF4 -. TBA + likely also form ion pairs with the 2:1 assembly which is shown by a slight downfield shift of Hb at higher concentrations of TBABF4 ( Figure S50a  To verify that BF4binds inside the cyanostar cavity, a 19 F NMR titration monitoring the addition of TBABF4 to pCS was examined. The 19 F NMR titration shows medium exchange peaks ( Figure S50a), which differs from the fast exchange 1 H NMR peaks ( Figure S50a). Early in the titration, a single 19 F peak is observed at -148 ppm ( Figure S51a), which shows an inflection point at 0.5 equivalents TBABF4 ( Figure S51b). This peak is shifted 3 ppm downfield from uncomplexed TBABF4, which is consistent with the binding of BF4inside cyanostar. The inflection point at 0.5 equivalents is consistent with the formation of a 2:1 complex with pCS. Upon further addition of TBABF4, an additional peak appears at 152 ppm ( Figure S51a) that grows in intensity ( Figure  S51c). This peak is similar to uncomplexed TBABF4 at 151 ppm, which is consistent with excess TBABF4 present is solution.

Orientation of the Macrocycles in the Dimer
The diffusion coefficients and ROESY correlations were examined to help distinguish one arrangement over the other (syn, meta and anti). The diffusion coefficients for the monomeric control [POPCu-CS] + , 5.5 ±0.2 × 10 −10 m 2 / s, and the target complex [POPCu-CS-BF4-CS-CuPOP] + from the one-pot assembly, 4.8 ±0.6 × 10 −10 m 2 / s, are the same within error. The diffusion coefficients were the same whether accessed from either cation-anion, 4.6 ±0.4 × 10 −10 m 2 / s, or anion-cation pathways, 4.9 ±0.2 × 10 −10 m 2 / s, and they are barely smaller than the monomeric control. As a result, the ratio of [POPCu-CS-BF4-CS-CuPOP] + and [POPCu-CS] + ranges 1.0-1.4. The ratios of the three conformations relative to the control and as calculated from molecular modelling are 1.0, 1.2, and 1.4 for the syn, meta and anti, respectively. As a result, diffusion NMR cannot be used to determine the conformation, but the syn and meta isomers better match the experimental data. 1

Choice of Anions for the Dimerization of Cationic Intermediates and Follow-Up Studies to Investigate Viability of Sequence-Dependent Target Product
Our selection of anions was strategic. The BF4and ClO4anions bind strongly to cyanostar but are typically weakly coordinating with metal cations. We conducted additional studies using S54 dibenzyl phosphate ( Figure S71-S72, S75) and dibutyl phosphate ( Figure S73-S74) that each bind well to cyanostar 15 but can also coordinate to metal complexes. We hypothesized that these stronger anions would interact with the metal ion to disrupt the overall assembly when we used either a one-pot method or prepared the cationic intermediate. We also hypothesized that precomplexation of these anions with the macrocycle would prevent the phosphates from coordinating with the metal ions. To evaluate this idea, we pre-complexed these phosphate anions with cyanostar. For this purpose, we used control compounds: parent cyanostar (pCS), dibutyl phosphate, dibenzyl phosphate, and a model Cu + complex. However, we found that precomplexation failed to prevent the phosphate from coordination of the metal centers. We observed that the copper complexes decomposed in 24 hours ( Figure S71-S75).
We investigated the interactions of dibutyl phosphate (P − ) with Au + -based cysnostar complexes. We added P − to [PPh3Au-CS] + NTf2 − ( Figure S76, the middle spectrum) and expected to form [PPh3Au-CS-P-CS-AuPPh3] + . The phosphate peak (Hα) at 4.5 ppm ( Figure S76, inset) suggested that some phosphate anions form [3]pseudorotoxane in accordance with our previous report, 16 while the rest of the phosphate did not bind with cysnostar cavity. The broad signature of NMR peaks, especially in the aromatic region, did not allow us to characterize the spectrum to a well-defined assembly.
To deconvolute the spectrum, we did the following control experiments. We added P − to the model gold(I) complex ( Figure S77d). We saw a pronounced upfield shift of the imine proton (Ha), which suggested the detachment of gold ion from the model complex. We inferred that the ionic interaction between positively charged Au + and P − was stronger than the coordination of Au + with pyridyl-imine ligand. We also observed a small peak (marked as * in Figure S77d) corresponding to the aldehyde proton of F-PyCHO. So, the addition of P − to the model gold(I) complex resulted in a mixture of Ph3PAu + P − , imine, and its Schiff base precursors. To this mixture, we added pCS (2 equiv. with respect to P − ) ( Figure S77c). Our expectation was that if the binding of P − with pCS were higher than the ionic interactions between Ph3PAu + and P − , pCS would bind with P − to form a [3]pseudorotaxane, and the remaining Ph3PAu + would reform the model gold(I) complex. The NMR spectrum ( Figure S77c) showed that we formed the desired pseudorotaxane based on pCS and P − , but we did not reform the model gold(I) complex. The broad new peaks (marked as # in Figure S77c) indicated the ionic interaction of Ph3PAu + with P − , which was bound in the pseudorotaxane.
Taking all these control experiments together, we concluded that the addition dibutyl phosphate (P − ) to [PPh3Au-CS] + NTf2 − led to decomplexation of gold ion, resulting in a soup of different fragments ( Figure S78).

X-Ray Crystallography
Single crystals of gold(I) model complex suitable for X-ray diffraction were grown by vapor diffusion of diethyl ether into a dichloromethane solution of the complex. A yellow crystal ( Figure  S65, block, approximate dimensions 0.15 × 0.14 × 0.12 mm 3 ) was placed onto the tip of a MiTeGen pin and mounted on a Bruker Venture D8 diffractometer equipped with a PhotonIII detector at 100.0 K. Figure S80. Microscope images of bulk material and crystal selected.

Data collection
The data collection was carried out using Mo K radiation ( = 0.71073 Å, IS micro-source) with a frame time of 2 seconds and a detector distance of 40 mm. A collection strategy was calculated and complete data to a resolution of 0.70 Å with a redundancy of 4.5 were collected. The total exposure time was 1.36 hours. The frames were integrated with the Bruker SAINT 17 software package using a narrow-frame algorithm. The integration of the data using a triclinic unit cell yielded a total of 88532 reflections to a maximum θ angle of 30.52° (0.70 Å resolution), of which 10365 were independent (average redundancy 8.541, completeness = 99.7%, Rint = 3.16%, Rsig = 1.70%) and 9867 (95.20%) were greater than 2σ(F 2 ). The final cell constants of a = 8.6133(4) Å, b = 14.6051(6) Å, c = 15.5136(8) Å, α = 62.663(2)°, β = 79.227(2)°, γ = 82.934(2)°, volume = 1701.53(14) Å 3 , are based upon the refinement of the XYZ-centroids of 9286 reflections above 20 σ(I) with 5.298° < 2θ < 60.97°. Data were corrected for absorption effects using the Multi-Scan method (SADABS). 18 The ratio of minimum to maximum apparent transmission was 0.862. The calculated minimum and maximum transmission coefficients (based on crystal size) are 0.5470 S92 and 0.6100. Please refer to Table S10 for additional crystal and refinement information.

Structure solution and refinement
The space group P-1 was determined based on intensity statistics and systematic absences. The structure was solved using the SHELX suite of programs 19,20 and refined using full-matrix leastsquares on F 2 within the OLEX2 suite. 21 An intrinsic phasing solution was calculated, which provided most non-hydrogen atoms from the E-map. Full-matrix least squares / difference Fourier cycles were performed, which located the remaining non-hydrogen atoms. All non-hydrogen atoms were refined with anisotropic displacement parameters. The hydrogen atoms were placed in ideal positions and refined as riding atoms with relative isotropic displacement parameters. The final full matrix least squares refinement converged to R1 = 0.0213 and wR2 = 0.0514 (F 2 , all data). The goodness-of-fit was 1.030. On the basis of the final model, the calculated density was 1.861 g/cm 3 and F(000), 932 e -.

Structure description (CCDC Deposition Number 2091810)
Asymmetric unit (CF3SO2N − molecules have 0.5 occupancy each and are disordered over a special position)