Experimental Quantification of Halogen⋅⋅⋅Arene van der Waals Contacts

Abstract Crystallographic and computational studies suggest the occurrence of favourable interactions between polarizable arenes and halogen atoms. However, the systematic experimental quantification of halogen⋅⋅⋅arene interactions in solution has been hindered by the large variance in the steric demands of the halogens. Here we have synthesized molecular balances to quantify halogen⋅⋅⋅arene contacts in 17 solvents and solvent mixtures using 1H NMR spectroscopy. Calculations indicate that favourable halogen⋅⋅⋅arene interactions arise from London dispersion in the gas phase. In contrast, comparison of our experimental measurements with partitioned SAPT0 energies indicate that dispersion is sufficiently attenuated by the solvent that the halogen⋅⋅⋅arene interaction trend was instead aligned with increasing exchange repulsion as the halogen increased in size (ΔGX⋅⋅⋅Ph=0 to +1.5 kJ mol−1). Halogen⋅⋅⋅arene contacts were slightly less disfavoured in solvents with higher solvophobicities and lower polarizabilities, but strikingly, were always less favoured than CH3⋅⋅⋅arene contacts (ΔGMe⋅⋅⋅Ph=0 to −1.4 kJ mol−1).

attractive and repulsive components of van der Waals interactions: London dispersion vs. steric repulsion. [9]Theoretical studies indicate that dispersion is the major stabilising component in halogen-arene interactions in the gas phase, while dispersion-correction is often required to predict interaction geometries observed in the solid state. [10]6a,11] The energetics of interactions in solution are further complicated by the very substantial, but incomplete cancellation of dispersion due to competing interactions with the surrounding solvent. [12]ere we present an experimental investigation of halogen•••arene interactions.A series of Wilcox molecular torsion balances were synthesised (Figure 1), [13] and thermodynamic double-mutant cycles (Figure 2) were used to dissect the energetics of halogen-arene interactions.The roles of solvophobic effects and competitive dispersion interactions were determined via experimental measurement in 17 solvent systems and by comparison with gas-phase computational data (Figures 3 and 4).
Previous experimental studies of halogen•••arene interactions have employed supramolecular complexes [14] and molecular balances. [15]Shimizu discovered stabilizing interactions between orthogonally orientated CÀ F bonds and aromatic rings, [15c] which mirrored Diederich's earlier unveiling of orthogonal fluorine•••amide interactions using Wilcox torsion balances. [13,16]Both studies determined orthogonal fluorine interactions to be primarily electrostatic and dipolar in origin.15b,d,f] Contrasting with prior studies in which halogen σ-holes are pointed towards aromatic π-clouds, [14,15c-d] we reasoned that Wilcox molecular torsion balances [12a,c,13,16,17a-b] could be adapted to instead examine near-parallel halogen•••arene contacts ((�)-1X and (�)-2X series in Figure 1).We anticipated this side-on geometry would be useful for investigating dispersion in halogen•••arene interactions, since the polarizable electron clouds are brought into contact, but perpendicular halogen bonds cannot form.3a][10] Even when a perpendicular halogen-bonding contact is made between a halogen σ-hole and an aromatic π cloud, the energetic contributions from electrostatics and electron delocalisation (aka polarisation, induction, donor-acceptor, orbital interactions) are smaller than the dispersion component. [10]inimized structures of the (�)-1X series of balances (calculated both with and without dispersion correction) consistently positioned the iodo, bromo, chloro, and methyl substituents in contact with the face of the terminal phenyl ring in the folded conformation (Figures 3 right, and Fig 2. Negative energies correspond to favourable X•••arene interactions.The example inset structures were calculated using M06-2X/def2-TZVP.Calculated interaction energies and geometries determined using other theory/ basis set combinations and X-ray structures [20] are provided in Tables S1-S6 and Figure

Communications
structures are influenced by crystal packing, they most closely resembled the DFT B3LYP minimizations.The crystal structures and B3LYP calculations featured longer halogen-arene distances than those determined using computational methods that include long-range correlation and dispersion corrections (M06-2X, ωB97X-D and ωB97X, Figure S3, Supporting Information).
The equilibrium folded/unfolded conformational ratio, K was determined for all 12 molecular balances in the (�)-1X and (�)-2X series (X=H, F, Cl, Br, I, Me) in 17 solvents and solvent mixtures using 1 H NMR spectroscopy at 300 K (Section S1, Supporting Information).The conformational free energy differences between the folded and unfolded conformers were determined from ~G = À RT lnK.All balances preferred the folded conformer in all solvents examined (Figure S1 and Tables S1-S2, Supporting Information).However, this conformational preference is not solely governed by the halogen•••arene interactions.12a,16,18] The dissected experimental ΔG X•••Ph values shown in Figure 3A reveal small interaction energies and weak solvent effects.All halogen substituents formed weaker interactions with the terminal phenyl ring (ΔG X•••Ph = 0 to + 1 kJ mol À 1 ) relative to the case where X=H.The I•••Ph interaction was generally less favored than the F•••Ph interaction.However, the F•••Ph interaction was preferentially weakened by H-bond acceptors such as pyridine-d 5 , DMSO-d 6 , and THF-d 8 /D 2 O, presumably due to the edges of the fluorophenyl rings being better H-bond donors than those bearing less electronegative halogens.In stark contrast to the disfavoured halogen•••arene interactions, the Me•••Ph interactions were universally favoured (ΔG Me•••Ph = 0 to À 1.4 kJ mol À 1 ).13b] Although it is possible that solvent effects might be cancelled out by the double-mutant cycle analysis, the ΔG values of the (�)-1X series taken in isolation reveal the same major trends: there is little variation as the halogen is changed, while the balance hosting the Me•••Ph interaction consistently presents the strongest preference for the folded conformation (Figure S1, Supporting Information).
The experimental interaction trends were reproduced via double-mutant cycle analysis of the calculated energies of the minimized structures at several levels of theory, even when lower levels of theory and small basis sets were used (Figure 3B and Figure S2, Supporting Information).SAPT0 energy decomposition calculations provided further insight into the origins of the energetic trends (Figure 4 and Section S2.2, Supporting Information). [17]The decreasing stability of the halogen•••arene interactions as the halogens increase in size is mirrored in only the steric (exchange) component.Interestingly, the Me•••Ph interaction has similar van der Waals components (dispersion and exchange) as the I•••Ph interaction, but a more favourable electrostatic con-tribution.These dominant electrostatic and steric effects are also consistent with the limited solvent effects seen in Figure 3A and reproduced in the gas-phase computational results in Figures 3B and S2, Supporting Information.
Gas-phase computed energies may be useful for understanding energetic trends across a series of closely related compounds, but the accurate calculation of solvent effects remains an unsolved challenge.In contrast, even subtle solvent effects can be measured experimentally using molecular balances.Initially, we sought to examine solvophobic influences by the addition of D 2 O to deuterated tetrahydrofuran (THF-d 8 , Figure 3A, bottom).Upon the addition of D 2 O, the Me•••Ph and I•••Ph interactions were more stabilized than those involving smaller halogens.Similarly, the most disfavoured halogen•••Ph interactions were found in less solvophobic solvents such benzene, CS 2 , and CCl 4 .However, the solvophobic effect was insufficient to make any of the halogen•••Ph interactions favourable (up to 40 % v/v D 2 O), and there was only limited evidence of the larger (more solvophobic) I•••Ph contact being slightly less disfavoured than those involving smaller halogens.Moreover, these energetic trends cannot be wholly attributed solvophobic effects since they could also arise from changes in competitive dispersion interactions with the solvent; increasing water content decreases bulk polarizability, while less solvophobic apolar solvents tend to have higher bulk polarizabilities (e.g.12c,19] Hence, we next examined whether halogen•••Ph interactions might be weakened by increasing solvent competition for dispersion interactions by increasing the bulk polarizability of the solvent. Counter to our expectations, adding the very polarizable molecule iodine to the most polarizable solvent, CS 2 tended to push the folding free energies of all balances closer to zero rather than making the halogen•••Ph interactions even less disfavoured (Figure 3A and Figure S1 in the Supporting Information).It is likely that adding iodine not only increased the bulk polarizability, but also increased solvent cohesion and hence the solvophobic effect; after all, iodine is a solid with a melting point of 114 °C.Nonetheless, it could still be seen that changes in the interaction energies upon varying the halogen substituent were largest in the most polarizable CS 2 /I 2 mixtures, and smallest in the least polarizable solvent examined (THF/40 % D 2 O).
In summary, we have synthesized molecular torsion balances to measure halogen•••arene interactions in a wide range of solvents.The adoption of the Wilcox balance framework and its inherent flexibility enabled the variable steric demands of the halogens to be accommodated.This enabled us to sidestep steric issues that have been widely encountered in previous experimental attempts to investigate interactions involving halogens.Side-on halogen•••arene contacts were found to be weakly disfavoured in all 17 solvents examined (0 to 1.5 kJ mol À 1 ).In stark contrast, identically positioned methyl•••arene interactions were weakly favourable in all solvents examined (0 to À 1.4 kJ mol À 1 ).SAPT energy decomposition calculations support the hypothesis that the CH 3 •••arene interactions were favoured over the halogen•••arene interactions due to Angewandte Chemie Communications electrostatic attraction, which is greatly diminished in halogen•••arene interactions.The halogen•••arene interactions, particularly those involving the larger halogens, were more sensitive than CH 3 -arene interactions to increasing the solvophobic effect via the addition of water.However, the solvophobic effect could not be increased to sufficient extent by the addition of up to 40 % v/v D 2 O in THF-d 8 to make any halogen•••arene interaction favourable.The main finding is that dispersion interactions alone are insufficient to drive the association of halogen•••arene contacts in solution, even between functional groups considered to be highly polarizable.This investigation studied side-on halogen•••arene contacts; it is likely that the energetics of orthogonal halogen•••arene contacts capable of forming halogen-bonds via σ-hole interactions will differ from those of the present study.

Figure 1 .
Figure 1.Wilcox molecular torsion balances synthesized for experimentally quantifying the energetics of "side-on" halogen•••arene van der Waals contacts.Conformational equilibrium constants, K were determined using 1 H NMR spectroscopy.
ure S3, Supporting Information).Having confirmed that the Wilcox balance framework was sufficiently flexible to accommodate halogen•••arene contacts and the varied steric demands of the halogens, the (�)-1X and (�)-2X series of balances shown in Figure1were synthesized (Section S3, Supporting Information).X-ray crystal structures were determined for all six balances in the (�)-1X series.All compound crystallized in the folded conformation (FigureS3, Supporting Information, CCDC deposition numbers: 2244208-2244213).Although the X•••Ph interaction geometries in the crystal

Figure 2 .
Figure 2. Thermodynamic double mutant cycle used to dissect the contribution of the X•••arene interactions to the conformational preference for the folded conformation in balance series(�)-1X.This thermodynamic dissection was applied to both the experimental and computational data depicted in Figure 3.