Bent and Twisted: Synthesis of an Alkoxy-Substituted (1,5)Naphthalene-paracyclophanediene

This contribution describes the synthesis of [2.2](1,5)naphthalenoparacyclophane-1,13-diene in four steps from 1,5-bis(bromomethyl)naphthalene and 1,4-benzenedimethanethiol. Consisting of 2,6-dioctyloxynaphthalene and benzene moieties, the effects of differing arene size on the structure, strain energy, and chemical reactivity of the cyclophanediene are examined. Despite a strain energy of 24.3 kcal/mol, the naphthalenoparacyclophanediene was unreactive toward a library of olefin metathesis catalysts. This diminished reactivity can be explained by the steric hindrance of the twisted olefin. Incorporation of an electron donor (naphthalene) into the rigid paracyclophanediene structure can allow for applications in optoelectronics, chiral ligands, and planar chiral materials.


Materials and Methods
All chemicals were purchased from Oakwood Chemicals, TCI Chemicals, Strem, Ambeed, or Millipore Sigma and used as received unless otherwise indicated. Xylenes was dried via 4 Å molecular sieves prior to usage. All reactions were carried out under ambient conditions unless otherwise noted. Flash column chromatography was performed using silica gel 60 Å (230-400 mesh) from Sorbent Technologies.
NMR spectroscopy characterizations were conducted at 25 ˚C on a Bruker Avance 400 MHz, 500 MHz, or 600 MHz spectrometers. Chemical shifts are reported in ppm and referenced to solvent residual peaks. Splitting patterns are reported as singlet (s), doublet (d), doublet of doublets (dd), triplet (t), quartet (q) and multiplet (m).
Mass spectra of samples in methanol were acquired with an Agilent 6224 Accurate-Mass TOF/LC/MS Spectrometer using an ESI ion-source.
For Chiral HPLC, the samples were separated and analyzed by an Agilent 1260 Infinity HPLC equipped with CHIRALPAK IA-3 column or OD-H01 column.
Absorption spectra were obtained using a Cary 100 UV-VIS Spectrophotometer by Agilent Technologies. Fluorescence spectra were collected on a QuantaMaster 40 Photon Technology International spectrofluorometer equipped with Xenon lamp source, emission and excitation monochromators, excitation correction unit, and PMT detector. All measurements were conducted at 25.0 ± 0.1 °C maintained by a Quantum Northwest cuvette temperature controller. Emission and excitation spectra were corrected for the wavelength-dependent response and wavelength dependent lamp intensity.

Chiral HPLC
General procedure: 8 mg of 6 and 9 were dissolved in separate vials in a 1:1 mixture of hexanes and isopropanol.

Screening Reactions With Olefin Metathesis Catalysts
General procedure for reaction screenings of 9 with olefin metathesis catalysts: In a nitrogen filled glovebox a stock solution of the desired initiator (10 mol%) was prepared in anhydrous, degassed THF or toluene. 9 (20 mg) was weighted out into a one-dram vial and brought into the glovebox, dissolved in anhydrous, degassed THF or toluene and transferred to an oven dried Schlenk tube. An appropriate amount of the catalyst solution was added for [9] = 100 mM. The Schlenk tube was sealed and removed from the glove box where it was subsequently wrapped in aluminum foil and kept at room temperature or placed in an oil bath at 50 ˚C or 100 ˚C and stirred. The reaction was cooled to room temperature and a large excess of deoxygenated ethyl vinyl ether (0.80 mL) was added and allowed to stir at room temperature for 12 hours. The reaction mixture was than concentrated down, allowed to dry under vacuum for one hour after which a 1 H NMR spectrum was recorded. Starting material was recovered by column chromatography. Scheme S1. Olefin Metathesis Catalysts screened.

In situ 1 H NMR experiments
Compound 9 as an inseparable mixture of enantiomers (30 mg, 0.041 mmol) and 12 or 15 were individually weighted out into 1 dram vials and brought into a nitrogen filled glovebox. 9 and the desired catalyst were separately dissolved in THF-d8 or toluene-d8 and combined for total volume of 0.418 mL ( [9] = 100 mM) and transferred into a J-Young NMR tube that was sealed. The sample was removed from the glovebox, wrapped in aluminum foil, and placed in an ice bath. The sample was removed from the aluminum foil and placed into a 600 MHz NMR set to 25 ˚C from t = 0. For the sample with 12, the spectrometer was then heated to 50 ˚C. 1 H NMR spectra were recorded every five minutes for the first hour and then every 20 minutes for 24 hours.

Benchmarking
To accurately model strain energies (∆Gring strain) of cyclophanes with density functional theory, benchmarking of basis sets and functionals were performed. Etheneolysis of 2 was modeled because the heat of etheneolysis, ∆Getheneolysis, is expected to correspond very closely to the ring strain such that: ∆Gring strain = -∆Getheneolysis The ring strain for 2 has experimentally been determined to be 42.0 kcal/mol. [1] Scheme S2. Ethenolysis reaction used to model the strain energy of 2 Functional Screen: Etheneolysis calculations were performed with Gaussian 16 (revision A.03) [2] at the DFT level of theory. Structures were constructed in Avogadro [3] and were pre-optimized using the UFF force field. [4] XYZ Coordinates of these structures were optimized in Gaussian using different functionals and the 6-31G* basis set [5][6] at a singlet spin state. The energies of these structures were calculated using single point calculations with the same functional and the 6-311G* basis set. These single point energies were used to determine the ∆Getheneolysis via the following equation: Where E is the single point energy of the optimized structure. The functionals that were screened were: M05, M06, M08HX, M11, MN15, B3LYP, O3LYP, X3LYP, B3P86, BH&HLYP, PW6B95, BMK, N12, TPSSh, ωB97X-D, LC-ωHPBE, APF, and APF-D. From these calculations, we obtained a spread of strain energies (Table S2, Figure S10).  a basis set screen was performed with 6-311G, 6-311G*, aug-cc-pVDZ, cc-pVDZ, aug-cc-pVTZ, cc-pVTZ, def2-SVP, def2-TZV, def2-TZVP,  def2-TZVPP and MidiX. From these calculations, we obtained a spread of strain energies (Table  S3, Figure S11). Figure S11. Ring strain energies (yellow bars) and difference from accepted value (inset, grey bars) for 2 at different functionalsthe experimental value of 42 kcal/mol is marked as a red horizontal line.

ring-strain with various basis sets w/ M11
S14 Based on these ∆Gring strain values, M11/6-311G* level of theory is the level of theory that is most able to reproduce the experimentally determined 42.0 kcal/mol strain energy of 2. As such, we proceeded with using M11/6-311G* as the level of theory for energy calculations.

Etheneolysis Calculations
Etheneolysis calculations were performed with Gaussian 16 (revision A.03) [2] at the DFT level of theory. Structures were constructed in Avogadro [3] and were pre-optimized using the UFF force field. [4] XYZ Coordinates of these structures were optimized in Gaussian using the M11 functional [10] and the 6-31G* basis set [5][6] at a singlet spin state. Optimized structures were confirmed from frequency calculations at the same level of theory, with all molecules described having no negative-valued vibrational modes. Energies of the optimized molecules were evaluated using single-point calculations at the M11/6-311G* level of theory. Finally, thermodynamic corrections were applied at 298 K using the entropy and enthalpy values obtained from the frequency calculations. The ring strain energy, ∆Gring strain, was taken as the negative value of the heat of etheneolysis.
For compound 9, etheneolysis of a truncated version (methoxy side-chains, not octyloxy) of the molecule, 17, was used to determine the ring strain (Scheme S3).
Scheme S3. Etheneolysis reaction used to determine the ring strain of 17. The strain energy of 17 is substantially lower than the strain energies of other cyclophanedienes such as 2, [5] and 19 and is about the same as the strain energy of norbornene (24.7 kcal/mol). [6]

Strain Energy Decomposition Calculations
To evaluate the total strain in 17 and understand the unusually low ∆Gring strain of 17, we utilized an approach reported by Grimme [10] to decompose the cyclophane strain energy into meaningful components. Calculations were performed with Gaussian 16 (revision A.03) [2] at the DFT level of theory. Structures were constructed in Avogadro and pre-optimized in Avogadro using the UFF force field. XYZ Coordinates of these structures were optimized in Gaussian using the M11 functional and the 6-31G* basis set at a singlet spin state. Frequency calculations were not performed due to the presence of structures along the thermodynamic cycle used for energy decomposition analysis that do not correspond to equilibrium geometries (therefore, vibrational analysis would yield meaningless results). Energies of the optimized molecules were evaluated using single-point calculations at the M11/6-311G* level of theory.
The energy decomposition analysis described breaks the total strain energy, ∆Etotal strain, into a sum of four components that arise from isodesmic reactions along a complete etheneolysis reaction path: -∆Etotal strain = ∆Etotal etheneolysis = ∆Edealkenylation + ∆Ealkenylation + ∆Eπ-interaction + ∆Earene strain Where ∆Edealkenylation corresponds to the energy of replacing the bridging alkenes with H atoms (but keeping the arenes in the same geometry), ∆Eπ-interaction corresponds to the energy of separating the two bent arenes, ∆Earene strain corresponds to the energy of relaxing the arenes from their bent geometries to their flat, equilibrium geometries, and ∆Ealkenylation corresponds to the energy penalty of replacing the alkenes (yielding complete etheneolysis products). The reactions used to calculate these energies for 17 are depicted in Scheme S5.
Scheme S5. Thermodynamic cycle corresponding to energy decomposition of ∆Etotal strain of 17.

S16
To fully understand the values of ∆Edealkenylation, ∆Ealkenylation, ∆Eπ-interaction, and ∆Earene strain of 17 in the context of other cyclophanes, we also calculated these values for compound 19 via a similar cycle (Scheme S6). The values obtained from both thermodynamic cycles are in Table S4.
Scheme S6. Thermodynamic cycle corresponding to energy decomposition of ∆Etotal strain of 17. *Note: although 30 is visually similar to 23, the 3-dimensional geometries of 30 and 23 are different because they arise from different cyclophanes. The ∆Etotal strain is 14.9 kcal/mol smaller in 17 than in 19. While this result was surprising given the size mismatch of the arenes in 17 and the highly deformed naphthalene, the energy decomposition helps to rationalize this result. First, the ∆Edealkyenylation is 18.0 kcal/mol smaller in 17 than in 19, meaning the transformation from 17 to 21 is less favorable than the transformation of 19 to 29 (Scheme S7).

Scheme S7. The dealkenylation steps for energy decomposition analysis of A) 17 and B) 19.
This difference in ∆Edealkyenylation is rationalized from a structural perspective: in 19, the bridging alkenes make a 90º dihedral angle with the place of the benzenes, preventing any π-conjugation from occurring. In 17, the alkene bridges make a 65º dihedral angle ( Figure S12), enabling some amount of stabilizing π-conjugation between the alkene bridges and arenes, making ∆Edealkyenylation less favorable in 17 and reducing the total strain. Figure S12. Single crystal XRD structure of 9, oriented to highlight the ~65º dihedral angle between the arenes and alkene bridges.
Next, the ∆Eπ-interaction is more negative for 17 than 19, indicating that separating the two arenes in structure 21 is more favorable than in structure 29 (Scheme S8).

Scheme S8. The π-interaction evaluation step for energy decomposition analysis of A) 17 and B) 19.
Given that structures in 17 and 19 are donor/acceptor systems, the more favorable ∆Eπ-interaction in 17 is rationalized on the basis of the arenes in 21 are slightly closer (dcentroid-to-centroid(17) = 2.948 Å) than the arenes in 19 (dcentroid-to-centroid(19) = 2.981 Å) causing slightly more unfavorable πinteractions in 21 than in 29 and overall adding slightly to the ring strain of 17 compared to 19.
Next, the ∆Earene strain is 7.0 kcal/mol more negative for 17 than 19, indicating that in sum the arenes are more strained in 17 (Scheme S8). This is likely due to the large size mismatch between the benzene and the naphthalene in 17, evidenced by the H-to-H distances (dHH) in the equilibrium geometries of 25 (dHH = 4.96 Å) and 26 (dHH = 5.59 Å) with a ∆dHH = 0.63 Å, while the sizes of the two arenes in 19 are well-matched, with a ∆dHH = 0.01 Å because both are benzene. This mismatch in size contributes to the ring strain of 17 compared to 19.
Scheme S9. The arene-strain evaluation step for energy decomposition S18 analysis of A) 17 and B) 19.
Lastly, the ∆Ealkenylation is 4.46 kcal/mol more positive in 17 than in 19, indicating that alkenylation is less favorable in 17 than 19.
Scheme S10. The alkenylation step for energy decomposition analysis of A) 17 and B) 19.
We ascribe this to the unfavorable steric interactions present in the product 28 due to the substitution pattern of the naphthalene. The torsional angle between the arene and alkenes in 28 is approximately 38º ( Figure S13). In 27 and 33, the same torsional angle between the arene and alkene is nearly 0º. This deviation from planarity in 28 makes the alkenylation step in 17 less favorable because it reduces the conjugation between the arene an alkene. As such, the ∆Ealkenylation for 17 is more positive than in 19, reducing the driving force for ring-opening in 17 and therefore reducing the strain energy in 17. Figure S13. Top-down and side view of 28, highlighting the non-planar geometry caused by substituent steric repulsion.
Combined, these results demonstrate that the largest contributor to the reduced strain of 17 compared to 19 is the twisting of the rings that enables conjugation between the olefins and arenes, stabilizing 17. This is evidenced by the less negative ∆Edealkenylation in 17 compared to 19. The effect of mismatched arene sized was minimal from a strain-energy perspective, given that the difference in arene strain caused by the mismatched size (7.0 kcal/mol additional strain) was nearly as large as the effect of steric repulsion between substituents in the product (5.4 kcal/mol less strain). In arenes with different substitution patterns (in which the products do not have torsional strain, like the strain present in 28) we expect the strain energy can be increased relative to 17. It is unlikely, however, that other cyclophanedienes with size mis-matched arenes will have larger strain energy than cyclophanedienes in which both arenes are benzene, because the size-mismatch between the arenes will contribute to twisting of the rings which reduces the dihedral angle between the olefins and arenes, reducing the strain energy.

Calculations with Ruthenium Catalysts
In the main text, we noted that 9 is unable to undergo ring-opening metathesis polymerization with olefin metathesis catalysts. We rationalized that this is because the hydrogens at the C4 and C7 positions block the olefin metathesis catalyst from coordinating to the olefins of 9. To understand the steric environment of such olefin metathesis catalysts while π-bound to the cyclophanes, we simulated Grubbs' second-generation initiator coordinated to the olefins of 17. We also performed the same analysis with the catalyst bound to 19, because monomers with similar structures to 19 are easily polymerized via ROMP.
Calculations were performed with Gaussian 16 (revision A.03) at the DFT level of theory. Structures were constructed in Avogadro and pre-optimized in Avogadro using the UFF force field. XYZ Coordinates of these structures were optimized in Gaussian using the M11 functional and the 6-31G* basis set at a singlet spin state. The geometry and relevant distances are shown in Figures S14 and S15.  Taken together, we believe that the shorter Ru-H distance in 35 arises from steric blocking of the Ru by the H atoms on the naphthalene moiety in 9, leading to worse binding of Ru to the olefins and consequentially, longer Ru-C distances in 35 compared to 34. The weaker binding of the Ru catalyst to 9 likely prevents the formation of the metallacyclobutadiene needed to facilitate olefin metathesis (and subsequent polymerization). This hypothesis is consistent with the spectra discussed in SI section 6, Figure S7, in which new carbene resonance appear in 1 H NMR spectra after the addition of monomer to Grubbs catalysts, indicating binding of the Ru to the olefins of 9, but the new resonances quickly decrease in intensity as the Ru catalysts decomposes due to the impeded metallacyclobutane formation. Figure S16. The molecular structure of dithia[3.3]naphthalenoparacyclophane 6 (ellipsoids set at 50% probability).

X-Ray Crystallographic Data
A colorless, block-like specimen of C36H50O2S2, approximate dimensions 0.360 mm x 0.500 mm x 0.540 mm, was used for the X-ray crystallographic analysis. The X-ray intensity data were measured on a Bruker D8 SMART APEXII three-circle diffractometer system equipped with a Incotec microfocus sealed X-ray tube (MoKα, λ = 0.71073 Å) and a multilayer optics monochromator.
A total of 1620 frames were collected. The total exposure time was 2.06 hours. The frames were integrated with the Bruker SAINT software package using a narrow-frame algorithm. The integration of the data using a monoclinic unit cell yielded a total of 57951 reflections to a maximum θ angle of 28.34° (0.75Å resolution), of which 8030 were independent (average redundancy 7.217, completeness = 99.8%, Rint = 4.95%, Rsig = 3.17%) and 6656 (82.89%) were greater than 2σ( The structure was solved and refined using the Bruker SHELXTL Software Package, using the space group P 1 21/c 1, with Z = 4 for the formula unit, C36H50O2S2. The final anisotropic fullmatrix least-squares refinement on F 2 with 363 variables converged at R1 = 3.80%, for the observed data and wR2 = 10.17% for all data. The goodness-of-fit was 1.084. The largest peak in the final difference electron density synthesis was 0.341 e -/Å 3 and the largest hole was -0.340 e -/Å 3 with an RMS deviation of 0.049 e -/Å 3 . On the basis of the final model, the calculated density was 1.191 g/cm 3 and F(000), 1256e -.  Figure S17. The molecular structure of [2.2]naphthalenoparacyclophanediene 9 (ellipsoids set at 50% probability).

S23
A colorless, block-like specimen of C36H46O2, approximate dimensions 0.300 mm x 0.370 mm x 0.560 mm, was used for the X-ray crystallographic analysis. The X-ray intensity data were measured on a Bruker D8 SMART APEXII three-circle diffractometer system equipped with a Incotec microfocus sealed X-ray tube (MoKα, λ = 0.71073 Å) and a multilayer optics monochromator.
A total of 2609 frames were collected. The total exposure time was 5.46 hours. The frames were integrated with the Bruker SAINT software package using a narrow-frame algorithm. The integration of the data using a triclinic unit cell yielded a total of 41732 reflections to a maximum θ angle of 28.32° (0.75 Å resolution), of which 7257 were independent (average redundancy 5.751, completeness = 99.8%, Rint = 3.73%, Rsig = 2.91%) and 5916 (81.52%) were greater than 2σ(F 2 ). The structure was solved and refined using the Bruker SHELXTL Software Package, using the space group P -1, with Z = 2 for the formula unit, C36H46O2. The final anisotropic full-matrix leastsquares refinement on F 2 with 410 variables converged at R1 = 4.64%, for the observed data and wR2 = 12.51% for all data. The goodness-of-fit was 1.064. The largest peak in the final difference electron density synthesis was 0.315 e -/Å 3 and the largest hole was -0.205 e -/Å 3 with an RMS deviation of 0.039 e -/Å 3 . On the basis of the final model, the calculated density was 1.161 g/cm 3 and F(000), 556 e -.

XYZ coordinates of Computed Structures
The format of the XYZ structures below is as follows: Chemdraw image First line = the number of atoms in the structure Second line = a comment line which contains the compound number and the computed electronic energy in hartree and any thermodynamic corrections Third line onward = the element, x-coordinate, y-coordinate, and z-coordinate, in Å. This document is constructed so that structures can be opened in molecular editors that can read .xyz files, such as MOLDEN or Avogadro.