Origin of the Enantioselectivity

To aid the understanding of the origin of enantioselectivity, molecular modeling was performed. Using (R,R,R,R)-configured 50e as the representative bis-hydrazone ligand, the corresponding palladium complex was computed in the following manner. The coordinates of the two chelating nitrogens and the two aromatic substituents on the palladium were approximated using ethane-1,2-dimine and phenyl groups by density-function theory (DFT) at

B3LYP/6-31G* level (Figure 22). This simplified structure was then decorated with (2R,5R)- diphenylpyrolidines and the two coupling moieties, 2-methyl-1-naphthyl and 1-naphthyl groups to furnish the diarylpalladium-bishydrazone complex (50e)PdR2. The core structure surrounding

palladium was kept constant while the remainder of the molecular was optimized by semi- empirical method at PM6 level.

Figure 22. Modeling of diarylpalladium complex of (R,R,R,R)-50e at ground state.

To clarify the spatial environment created by the ligand, the two naphthyl moieties are temporarily removed (Figure 23, left). Because of the C2-symmetry of the ligand, the NW and

SE quadrants are both occupied by the phenyl group from the ligand. On the contrary, empty spaces can be seen in the SW and NE quadrants. These chiral pockets would be expected to dictate the orientation of the substrates (Figure 23, right). In the current model, the two empty quadrants are occupied by the naphthyl rings and the relatively less hindered Me-C(2) and H- C(2) substituents are positioned in the more congested quadrants. The two naphthyl groups are not aligned but are tilted in a propeller sense. Because of the presence of these aromatic groups, the pyrrolidine rings of the ligand are slightly distorted away from the plane containing the bis- imine to alleviate steric strain.

To facilitate discussion, a simplified picture is provided to depict the steric environment created by the ligand and the position of the naphthyl group. The shaded block represents filled quadrant and the label B represents the location of naphthyl B-ring.

Figure 23. Steric environment imposed by chiral ligand (R,R,R,R)-50e.

On the basis of this model, the origin of enantioselectivity can be rationalized. The transition structure leading to the C-C bond forming event is likely accompanied by a clockwise conrotatory action of both naphthyl groups to avoid head-on collision between Me-C(2) and H- C(8) (Figure 24). This movement leads to the expected (S)-product from a (R,R,R,R)-bis- hydrazone ligand. The counterclockwise motion would engender unfavorable steric interaction between the 2-methyl substituent and phenyl group of the ligand. This motion leads to (R)- product. If the enantioselectivity observed for this coupling (95:5) is contributed by these opposing movements, an energy difference of 1.9 kcal/mol at 70 °C can be calculated between the two respective transition structures. The above hypothesis implies that the energy difference between TS-S and TS-R would be greater if both naphthyl moieties have a methyl substituent at the 2-position, because two unfavorable steric interactions are engendered when both C−Pd bonds are rotated in the counterclockwise direction. Consistent with this hypothesis, an energy difference of 3.0 kcal/mol was estimated based on 92:8 er at 110 °C for the coupling of potassium 2-methyl-1-naphthyldimethylsilanolate and 1-bromo-2-methylnaphthalene.

NW NE SW SE A B B A 1 2 3 4

Pd G‡_{= G}‡ R GS =1.9 kcal/mol ‡ G‡S G‡R PR(minor) PS(major) backward backward TS(S) TS(R) C' C front view C' Ligand Environment (S) Me C N _{N N} N Ph Ph Ph Ph 95:5 er (R,R,R,R)-50e 35

Figure 24. Hypothesis for the origin of enantioselectivity.

This model also suggests that a less sterically encumbered substituent such as a methoxy group at the 2-position of the substrate, should lower the transition state energy leading to the minor (R)-product (Figure 25). Consequently the ∆∆Gǂ between TS-S and TS-R is decreased and a lower enantioselectivity is expected. At 70 °C, the 2-methoxybinaphthalene was obtained in 88:12 er which is lower than 95:5 er obtained for 2-methylbinaphelene at the same temperature.

Figure 25. Hypothesis for decreased enantioselectivity for the coupling of less hindered substrate.

Applying the same logic, the lower enantiomeric ratios observed using 3,5-disubsituted aryl ligands may be understood (Figure 26). These substituents extend their presence into the empty quadrants. As a result, the naphthyl groups encounter some steric repulsion not experienced with the phenyl substituted ligand 50e. This interaction is indicated by the double- headed arrows in both front view and side view. On the contrary, the 3,5-subsitituents are distal to the 2-methyl group of the substrate. Therefore, the transition-state energy corresponding to counterclockwise rotation (TS-R) is relatively unaffected. The overall effect is a decreased ∆∆Gǂ which is reflected by a lower er.

Figure 26. Hypothesis for decreased enantioselectivity using 3,5-disubstituted aryl hydrazone ligands.

These lines of thought suggest that higher enantioselectivity may be possible if steric interaction can be introduced to disfavor counterclockwise rotation. In this regard, the 2-tolyl substituted ligand 50e was proposed because of the directionality of the 2-methyl substituent pointing directly toward 2-methyl group of the substrate. Furthermore, the 2-methyl group of the ligand has the opportunity to rotate away from NE and SW quadrants to create empty pockets similar to that of the phenyl ligand 50e-palladium complex. Unfortunately, racemic coupling product was observed which may suggest free rotation of the C(2)−tolyl bond that creates steric

repulsion at all four quadrants. Alternatively, this ligand may simply be too bulky for effective coordination resulting in background reaction.

Although adequate rationalizations can be provided for a number of observation described above, the crude model cannot justify the lower enantioselectivity observed using electron-deficient ligand 50c and 50d solely on the basis of steric arguments. As has been hinted at by Fernández et. al., arene-arene interaction may favor alternative orientation of the two naphthyl groups with respect to the ligand such that the B-rings are in the same quadrants as phenyl group. With this consideration, three more limiting diastereomeric diarylpalladium complexes can be formulated (Figure 27); one of which has two potential arene-arene interactions. These isomers are expected to have a preference for one of the two enantiomeric biaryl products.

Figure 27. Numbers of potential arene-arene interaction for diarylpalldium complexes.

In collaboration with Houk and Liu at UCLA, these diastereomeric complexes were subjected to calculation at a high level of theory to account for the solvent effect and dispersion interaction. The geometry was optimized with density function theory at B3LYP level with a mixed bases set; SDD for Pd and 6-31G(d) for other atoms. Single point calculations were performed at M06 level with SMD solvent model and a larger basis set, SDD for Pd, and 6- 311+G(d,p). A total of 8 transition structures (Figure 28) were located for the diarylpalladium complex; half of which simulate clockwise conrotatory reductive elimination (TS-A, TS-B’, TS- C and TS-D’) and the other half simulate counterclockwise conrotatory reductive elimination (TS-A’, TS-B, TS-C’, TS-D).

Figure 28. Transition structures for diarylpalladium complex of phenyl substituted bis- hydrazone ligand (R,R,R,R)-50e. Unfavorable interactions between the two naphthyl moieties are indicated by the distances between two atoms (< 80% of the sum of van der Wall radii).

In accord to the low-level ground-state calculation for A discussed previously, the most stable transition structure TS-A also features a propeller arrangement pre-disposed to the formation of biaryl (S)-35. The B-rings of the naphthyl groups are located at the relatively empty NE and SW quadrants. The transition structure, TS-A’, leading to the (R)-product has

TS-A ∆∆G‡_{= 0.0 kcal/mol} (S)-product TS-B ∆∆G‡_{= 2.9 kcal/mol} (R)-product TS-A’ ∆∆G‡_{= 7.3 kcal/mol} (R)-Product TS-B’ ∆∆G‡_{= 8.0 kcal/mol} (S)-product 2.20 TS-C ∆∆G‡_{= 3.9 kcal/mol} (S)-product TS-C’ ∆∆G‡_{= 5.1 kcal/mol} (R)-product 2.23 TS-D ∆∆G‡_{= 3.5 kcal/mol} (R)-product TS-D’ ∆∆G‡_{= 7.7 kcal/mol} (S)-product 4.27 B B

significantly higher energy because of steric repulsion between Me-C(2)/H-C(8’) (2.17Å) and H- C(8)/H-C(2’) (1.86 Å). The geometry of palladium is considerably distorted from square planar to minimize steric strain. Only one energy minimum was located for the ground-state structure for the two transition structures suggesting rapid interconversion between conformers from the rotation of Pd-CAryl bond at the ground-state.

The second most stable transition structure TS-B has two B-rings at the NW and SE quadrants. To minimize steric interaction between substrates and phenyl groups of the ligand, the palladium is again distorted from square-planar geometry. The (R)-product is expected from this diaryl-Pd arrangement. The clockwise conrotatory motion implied by TS-B’ engenders steric strain observed similarly for TS-A’; the distance between C(2)/H-C(8’) (2.26 Å) and between H- C(8)/H-C(2’) (1.86 Å) are within 80% of the sum of the van der Waal radii (Å).

The two naphthyl moieties in transition structures TS-C, TS-C’, TS-D and TS-D’ have syn relationship. These structures have unfavorable interaction between H-C(8)/C(8’) or C(8)/H- C(8’). Therefore, energies higher than those for TS-A and TS-B were found.

Based on the energy difference (2.9 kcal/mol) between the two most stable transition structures TS-A and TS-B, the predicted er (99:1) is comparable to observed er (95:5) at 70 °C.

The same calculations were applied to electron-deficient ligand 50c for the two lowest transition structures, TS-A and TS-B (Figure 29). This endeavor revealed a smaller energy difference between the two structures (1.4 kcal/mol), which may be attributed to arene-arene interaction between substrates (B-ring) and ligand (4-trifluorophenyl) in TS-B. The distances between the centroid of the two π-systems are 3.96 Å and 4.47 Å. The corresponding distances in the phenyl hydrazone-complex (R,R,R,R)-50e are 4.27 Å and 4.60 Å (Figure 28). These results bode well with the stronger arene-arene interaction between electron-rich and electron- deficient π-system than between two electron-rich π-systems.228 In either case, a longer distance is measured between the π-systems located at the NW quadrant than at the SE, because the 2- methyl substituent is repelled by the nascent naphthyl moiety. The computed er (91:9) based on the energy differencebetween TS-A and TS-B closely approximates the observed er (90:10) at 70 °C.

Figure 29. Transition structures for diarylpalladium complex of 4-trifluorophenyl substituted bis-hydrazone ligand (R,R,R,R)-50c.

3.5.7.1 Future Development

The modeling studies provided much insight into the origin and extent of enantioselectivity. Nevertheless, a number of issues remained to be addressed. First, the predicted enantioselectivity was based on the energy difference relative to the most stable transition structure such as TS-A. A more precise estimate of ∆∆Gǂ should be calculated from the respective ∆G values. Therefore, the ground-state energies corresponding to each transition structure, Aǂ-Dǂ need to be computed. Second, the validity of the advanced calculations need to be examined on other ligands such as 3,5-dimethyl and 3,5-bis(trifluoromethyl) substituted ligands. Lastly, the interconversion barriers between diarylpalladium complexes A-D need to be estimated. High energy barriers would suggest that reductive-elimination step does not determine the enantioselectivity and the product composition is a consequence of the ratio of A/B/C/D. Preliminary calculation at PM6 level indicates that interconversion is unlikely if the hydrazone ligand is bound to palladium in bidentate mode. The steric environment imposed by the ligand prohibits rotation of one naphthyl group from passing the other. However, interconversion may be possible if the ligand is bound in monodentate mode leaving palladium with an empty coordination site.229 This hypothesis is worth considering if the ligand does not associate to palladium tightly. Several observations suggest that bis-hydrazones are weakly coordinating ligands. Attempts to prepare oxidative addition complex by ligand displacement strategy with mono-, bidentate phosphine complexes and diamine complexes were unsuccessful. Inefficient

and incomplete displacement of cyclooctadiene from its PdCl2 complex was observed with

piperidine-based bis-hydrazone 46. The solution NMR of (S,S,S,S)-(50e)PdCl2 showed multiple

and undefined broad peaks at −40 °C, which could imply rapid equilibration between bidentate and monodentate binding modes. The allyl silyl ether byproduct generated from activation of [allylPdCl]2 was found to slightly lower the product er in some cases by acting as a competitive

ligand. These circumstantial evidences indicate conversion between diarylPd complexes A-D is a possibility through partial ligand dissociation. Importantly, the results from the donor/acceptor reversal experiments support reductive elimination as the stereodeteremining step implying that diastereomeric complexes A-D are in equilibration.

From a mechanistic point of view, the empty coordination site required for the transmetalation event necessitates the partial dissociation of the bidentate ligand (Scheme 143). The bulky tri-tert-butylphosphine ligand successfully employed in the preparative cross-coupling (Chapter 1) and Hammett study (Chapter 2) ensures a tricoordinate palladium intermediate to facilitate transmetalation. Bidentate phosphine ligands, a stronger chelating ligands than bis- hydrazones, were found to be generally less effective in the cross-coupling reaction of aryldimethylsilanolate. Transmetalation to a tetracoordinated palladium is unfavorable based on prior calculations.230 Therefore, an opportunity exists for the equilibration between A-D through tricoordinated palladium after transmetalation and before re-association of hydrazone ligand.

In summary, the refinement of the molecular model and enantiomeric selectivity predictions require further calculation of the ground-state energy for all diarylpalladium complexes and the interconversion barriers between them.