A ring of rotaxanes: studies of a large paramagnetic assembly in solution

Here we report the synthesis and structural characterization of four [7]rotaxanes formed by coordinating hybrid inorganic–organic [2]rotaxanes to a central {Ni12} core. X-ray single crystal diffraction demonstrate that [7]rotaxanes are formed, with a range of conformations in the crystal. Small angle X-ray scattering supported by molecular dynamic simulations demonstrates that the large molecules are stable in solution and also show that the conformers present in solution are not those found in the crystal. Pulsed EPR spectroscopy show that phase memory times for the {Cr7Ni} rings, which have been proposed as qubits, are reduced but not dramatically by the presence of the {Ni12} cage.


[2(1A) 6 ]
Compound 1A (75.0 mg, 0.03 mmol, 6 eq) was dissolved in 5 mL of toluene and added to a suspension of compound 2 (17.6 mg, 0.005 mmol, 1 eq) in 5 mL of toluene.The resulting mixture was stirred for 10 minutes at 60 °C before filtering the resulting solution.The solvent was removed under a reduced pressure to yield a green powder which was recrystallized from vapour diffusion of MeCN into toluene.Small, green, well-defined crystals suitable for characterisation by single crystal XRD were obtained (40.5 mg, 45 %).Elemental analysis (%) calcd.for C 684 H 1086 Cl 12 Cr 42 F 48 N 24 Ni 18 O 234 : C 45.72, H 6.09, N 1.87; found: C 45.86, H 6.13, N 1.51.

[2(1B) 6 ]
Compound 1B (75.0 mg, 0.08 mmol, 6 eq) was dissolved in 5 mL of toluene and added to a suspension of compound 2 (17.7 mg, 0.005 mmol, 1 eq) in 5 mL of toluene.The resulting mixture was stirred for 10 minutes at 60 °C before filtering the resulting solution.The solvent was removed under a reduced pressure to yield a green powder which was recrystallized from vapour diffusion of MeCN into toluene.Small, green, well-defined crystals suitable for characterisation by single crystal XRD were obtained 10 minutes at 60 °C before filtering the resulting solution.The solvent was removed under a reduced pressure to yield a green powder which was recrystallized from vapour diffusion of MeCN into toluene.Small, green, well-defined crystals suitable for characterisation by single crystal XRD were obtained (59.0 mg, 49 %).Elemental analysis (%) calcd.for C 714 H 1110 Cl 12 Cr 42 F 48 N 24 Ni 18 O 246 : C 46.25, H 6.03, N 1.81; found: C 46.42, H 5.97, N 1.76.
Images of structures and electron density maps were rendered using POV-Ray 3.6. 4

Refinement
Structure solution and refinement was carried out using Shelx-20XX, implemented through Olex2 v1.5.X2,X3 Least-squared refinements against all F 2 values using ShelXL 2019/3.In all cases, the chromium and nickel in the Cr7Ni rings were assumed to be equally distributed, with each metal site constrained to have a 7:1 ratio of chromium to nickel, with the associated same coordinate and same atomic displacement parameter constraints.All organic moieties (pivalate, 6-chloro-2hydroxopyridine, acetate and secondary ammonium cation threads) were refined to have similar 1,2and 1,3-bond distances.Where modelled, disordered moieties were refined over two positions with the occupancy refined against independent free variable such that the sum of occupancy for each disordered associated pair equals 100%.Global similar neighbour atomic displacement parameter and rigid bond restraints were applied to the model.All hydrogens were modelled in idealised calculated positions, and in the case of the methyl hydrogens in idealised staggered geometry without refinement of the torsion angles.Diffuse solvent voids were modelled using an applied calculated solvent mask to account for the solvent contribution to the observed intensities.
[2(1A) 6 ] Structure was solved using ShelXT 2018/2.In the case of the whole disordered Cr7Ni cluster (B/E), the inorganic chromium/nickel and fluoride rings were also refined to have similar 1,2-and 1,3-bond distances.There is evidence in the Ni12 cluster that there is substitution of 6-chloro-2-hydroxopyridine for acetate, and in one circumstance this was clear enough that this substitutional disorder could be satisfactorily modelled.
[2(1B) 6 ] Structure was solved using ShelXS.Additional fixed distance restraints were applied to the methylthioether group at the end of the ammonium cation threads.Flat restraints were also applied to the 6-chloro-2-hydroxopyridine moieties.Due to the low resolution of the data, all pivalate moieties were refined isotropically to retain a reasonable data to parameter ratio.
[2(1C) 6 ] Structure was solved using ShelXT 2018/2.There is evidence in the Ni 12 cluster that there is substitution of bridging acetate with pivalate, which was modelled as 100% pivalate.Additional 1,2and 1,3-same distance restraints were applied to all equivalent bonds across the 6-chloro-2hydroxopyridine moieties, rather than just the same bonds between moieties, in order to refine a suitable geometry.Flat restraints were applied to the terminal phenyl moieties and the neighbouring methylene carbon of the ammonium cation thread.
[2(1D) 6 ] Structure was solved using ShelXT 2018/2.Whole disorder of the central Ni12 cluster and bound ammonium cation threads (2D 6 ) was identified and modelled over two positions against a single occupancy free variable.Additional 1,2-and 1,3-same distance restraints were applied to all equivalent bonds across the disordered moieties, rather than just the same bonds between moieties, to refine a suitable geometry.Where necessary, fixed distance and flat restraints were also applied.Strong similar neighbour atomic displacement parameter and rigid bond restraints were applied to the disordered moiety.Due to the low resolution of the data, and extensive disorder of the 2D 6 moiety, all pivalate moieties were refined isotropically to retain a reasonable data to parameter ratio.Legacy restraints are also included in the instructions file from initial refinements of the disordered moieties as residues; these were not active during the final refinement steps.

Small angle X-ray scattering
SAXS measurements were performed on a HECUS SAXS/GISAXS instrument equipped with XENOCS micro focus CuK α (λ = 1.5407Å) source equipped with Montel optics and the diffracted X-rays collected with a Dectris Pilatus 100 K 2D detector.Samples were dissolved in toluene contained in borosilicate capillaries with diameter of 1 mm and wall thickness of 10 µm.Silver behenate was used for calibration of the instrument before every collection.Pure toluene collection was performed with identical conditions as the samples to allow consistent subtraction.Sample collections typically took 10000 seconds.All experimental data are the sum of the 2D radial distribution of the small angle X-ray diffraction converted to a 1D line graph.Irena SAS/SANS routines in Wavemetrics Igor Pro have been used for calibration, 5 data conversion and subsequent analysis.
The analysis involved subtracting the solvent contribution from the sample and solvent data before employing routines in Irena for the analysis.Pair distance distribution functions provided a reliable, simple and reproducible means for investigating the molecular sizes.The corrected data was analysed using the Moores method. 5Initially, the approximate size is determined and then function fitted to a region between large aggregate signals (small angles) and the statistically insignificant data at high angles.Fitting was repeated until a steady maximum size was achieved.

Atomistic molecular dynamics simulations
All MD simulations were performed at 298 K in a toluene solution to relax the most promising constructed structures, and to obtain more realistic ensemble structural averages from which the experimental SAXS and corresponding pair pair P(r) distribution is made are made comparison to.All simulations were performed using Gromacs 2016.4 molecular dynamics package 6,7 The parameterization of the systems was adapted to the General Amber Forcefield. 8A single atomistic structure of each conformer in a solution of toluene at in a cubic box size of 10.00 nm 3 was set up.
To prepare the systems prior to performing full production run simulation, energy minimisation was performed using the conjugate gradient algorithm.The minimised system was followed by a short NPT ensemble simulation for 10 ns (T = 298 K and p = 1.01325 bar) using the Nosé-Hoover thermostat 9 and the Parrinello-Rahman barostat. 10The particle-mesh Ewald method 11 was used to compute the electrostatics.Periodic boundary conditions were used and the time step for Integration was fixed at 1fs with the neighbour list updated every 10 fs.
Small Angle X-Ray Scattering (SAXS) was calculated on selected obviously different conformers of each rotaxane, subsequently the choice for the structure to be set up for Molecular Dynamics simulation to be compared directly to SAXS data from experiments.Scattering factors were taken from computed X-ray scattering factors from Hartree-Fock calculations. 12The SAXS box used for all calculated profiles was 100 nm with a X-ray wavelength of 0.154209 nm.SAXS profiles were plotted as a natural log of intensity ln I(q)(a.u.) against the scattering vector, q (Å).SAXS calculation were done on the whole structure alongside the various different rotated conformers.
Irena SAS/SANS routines in wavemetrics in Igor Pro 5 have been used to calculate the pair distribution function from the SAXS data obtained from MD.The Pair distribution function using routines in Irena has enables us to investigate and compare the molecular size to experimental data.Moores method 5  was used to approximate the size and then fitted to a function in the region of large aggregate signals (small angles) and the statistically insignificant data at high angles.The fitting was repeated for all data sets until a maximum size was achieved.
The SAXS profile and P(r) functions were all visually inspected and compared directly to corresponding experimental data.
To assess whether the disparity of the preliminary SAXS predictions was a result of disorder in the crystal structure, or possible detachment of the metal rings in solution, we have constructed and analysed the SAXS spectra of a comprehensive range of possible rotaxane model structures.
In total, 100 unique initial static conformations of the four rotaxanes (64 for [2(1A) 6 ], 4 for [2(1B) 6 ], 16 for [2(1C) 6 ] and 16 for [2(1D) 6 ]) were assessed for viability.These model structures were constructed from three components: the {Ni 12 } core ( 2), {Cr 7 Ni} anionic rings (1) and protonated threads (A, B, C and D).The initial structural conformations of the {Ni 12 } core and the {Cr 7 Ni} ring components were based on the XRD structures; however to facilitate construction of the full [7]rotaxane, each of the three main components was independently optimised in the gas phase using density functional theory (DFT) using the B3LYP exchange-correlation functional and a 6-31G(d) basis set on HCNOF atoms, and a LANL2 effective core potential/double zeta basis set for Cl, Cr(III) and Ni(II); all metals were high spin and ferromagnetically coupled within the 2 and anti-ferromagnetically coupled within the Cr 7 Ni rings.Since full optimisation of the model [7]rotaxane at this level was not possible, in order to aid selection of feasible models and initiate molecular dynamics simulations of the structures, a primitive distancebased scoring function was used to assess the number and severity of the bad-contacts (short nonbond interactions due to close proximity or overlap of functional groups) in each model.This was subsequently used to make minor rotational and translational adjustments to the relative orientations of the thread, metal ring, pivalate ligands and the {Ni 12 } core before simulation.
Figure S5 summarises the four threads and the rotational conformations of each thread used to construct the model rotaxanes.It may be noted that for [2(1A) 6 ], rotamers Q and R are 0.8 kJ mol -1 higher in energy than P and S at the B3LYP/6-31G(d) level, but for the other threads the rotamers are all energetically degenerate.In the XRD only conformers P were evident for [2(1A) 6 ], [2(1B) 6 ] and [2(1C) 6 ], and the Q conformer for [2(1D) 6 ] -nevertheless the resolution of the conformations of these structures was poor. Figure S10 illustrates how the meta-substituted pyridyl threads of [2(1A) 6 ] may be oriented in either the X or E-thread directions, and/or that different conformers may alternate around the {Ni 12 } ring; the symmetry of threads [2(1B) 6 ], [2(1C) 6 ] and [2(1D) 6 ] does limit the number of conformations since the X and E orientations are equivalent to P/Q or R/S rotamers.The nomenclature of the Newman projections defining the rotational conformers of the thread has been used to define the [7]rotaxane conformers as displayed in Figures S6 -S9.Each conformer contains coordinated threads where the pyridyl group binding alternates above and below the plane of 2. Each set of alternate threads is defined by a Newman projection which corresponds to the rotational conformer of the thread.For example, conformer PQ contains three coordinated threads containing rotamer P and three containing rotamer Q (Figure S7).S3).
Table S3.Angles measured on the SAXS conformers and crystal structures of [2(1A) 6, 33.5, 36.6, 34.9, 33.2, 37.0 The SAXS spectrum and its corresponding electron pair-pair distribution function P(r), were first determined theoretically for a selection of the static structures for each of the four [7]rotaxanes.The models selected are visually distinct conformers (Figures S6 -S9), constructed from the low energy conformations of each thread such that the final model was not too tightly packed to have a reasonable expectation of being formed.Whilst each of the thread conformations are likely to be found in solution, it should be noted that once coordinated to the {Ni 12 } core, and the anionic rings are threaded, the resulting structure has relatively little freedom to interconvert between conformers without dissociating to do so.
The results demonstrate that the entire structure appears to be stable in solution.We compared the experimental SAXS and pair distribution functions P(r) (PDF) (Figures S12 -15) with the predicted SAXS for an ensemble of structures obtained during the MD simulation and then examined how SAXS and PDF would vary through the rotated conformers.The best agreement with experimental data is found for the MD ensemble average.The other calculations are not consistent with the experimental observations as a static representation of each structure's electron distribution is significantly higher at particular distances signifying the extra peaks around the central averaged one seen experimentally and in MD.Various conformers selected illustrate the most obvious differences that can be seen when visually inspected.

EPR Spectroscopy
Q-band data were recorded on a Bruker E500 spectrometer equipped with an ER 5106 QT (Q-band, ca.34 GHz) resonator.
Continuous wave (CW) EPR spectra of [2(1A) 6 ], [2(1B) 6 ], [2(1C) 6 ] and [2(1D) 6 ] were measured at 5 K, both in the solid state and as a frozen solution (1 mM in dry 1:1 CH 2 Cl 2 :toluene).The spectra are dominated by the S = ½ ground states of the {Cr 7 Ni} rings (signal at ca. 13,750 G in Figure S16).The gtensor in all cases g xyz = 1.79, 1.79, 1.75.There is a broad feature at ca. 10000 G, which may be due to resonances from 2. The narrowness of the EPR signal indicates no significant magnetic interactions between 2 and 1X in any compound.Figure S17 shows a broad feature observed when measuring the powder CW EPR spectrum of [2(THF) 6 ] at 5 K.    S4.The values observed show no significant variation between compounds nor from isolated {Cr 7 Ni} rings.

*
Measurements made using major part of disordered crystal structure.

Figure
Figure S1.a. Single crystal XRD structure of [2(1A) 6 ].Colour scheme: Cr dark green, Ni lilac, O red, N blue, C grey, F yellow, Cl bright green.The Ni sites in the {Cr 7 Ni} rings are disordered over multiple positions.Disorder has been omitted for clarity.b.Electron density map of the asymmetric unit of [2(1A) 6 ].

Figure S2 .
Figure S2.Single crystal XRD structure of [2(1B) 6 ].Colour scheme as in Figure S1; S gold.The Ni sites in the {Cr 7 Ni} rings are disordered over multiple positions.Disorder has been omitted for clarity.b.Electron density map of the asymmetric unit of [2(1B) 6 ].

Figure
Figure S3.a. Single crystal XRD structure of [2(1C) 6 ].Colour scheme as in Figure S1.The Ni sites in the {Cr 7 Ni} rings are disordered over multiple positions.Disorder has been omitted for clarity.b.Electron density map of the asymmetric unit of [2(1C) 6 ]. 17

Figure
Figure S4.a. Single crystal XRD structure of [2(1D) 6 ].Colour scheme as in Figure S1.The Ni sites in the {Cr 7 Ni} rings are disordered over multiple positions.Disorder has been omitted for clarity.b.Electron density map of the asymmetric unit of [2(1D) 6 ].

Figure S6 .
Figure S6.Diagram displaying thread rotational conformers present in conformers PP and PT alongside images of conformers used to obtain SAXS spectra for molecules [2(1A) 6 ], [2(1B) 6 ], [2(1C) 6 ] and [2(1D) 6 ].Conformers PP and PT differ only in the direction of the ammonium thread with respect to the plane of 2. Identity of Newman projections are as defined in Figure S5.

Figure S10 .
Figure S10.The {Ni 12 } ring and six pyridine ligands which are substituted by the threads (all other species are omitted for clarity).X-and E-orientations are possible for the meta-substituent of the pyridine head-group in [2(1A) 6 ].Alternating pyridine groups (ball-and-stick or wireframe) can be substituted in [2(1B) 6 ] with different thread rotamers.

Figure
Figure S12.a, Pair pair distribution for various conformers of [2(1A) 6 ] (pink, blue, cyan, orange) alongside the crystal structure (bright green), experimental (black) and MD (red).b, MD versus experimental for clarity.Various conformers selected illustrate the most obvious differences that can be seen when visually inspected.

Figure
Figure S13.a, Pair pair distribution for various conformers of [2(1B) 6 ] (pink, blue) alongside the crystal structure (bright green), experimental (black) and MD (red).b, MD versus experimental for clarity.Various conformers selected illustrate the most obvious differences that can be seen when visually inspected.

Figure
Figure S14.a, Pair pair distribution for various conformers of [2(1C) 6 ] (pink, blue, cyan) alongside the crystal structure (bright green), experimental (black) and MD (red).b, MD versus experimental for clarity.Various conformers selected illustrate the most obvious differences that can be seen when visually inspected.

Figure
Figure S15.a, Pair pair distribution for various conformers of [2(1D) 6 ] (pink, blue) alongside the crystal structure (bright green), experimental (black) and MD (red).b, MD versus experimental for clarity.Various conformers selected illustrate the most obvious differences that can be seen when visually inspected.