The importance of the antiperiplanar effect in the nucleophilic addition of hydride to a carbonyl group.

Abstract

We have measured the relative rates for addition of lithium aluminum hydride, sodium borohydride and triethylsilane to the conformationally fixed ketone 44 and its $\alpha$-methyl, -methylthio, -methoxy, -chloro and -fluoro quasi-axial and quasi-equatorial derivatives 45-53. On the basis of a molecular mechanics study of the $\alpha$-chloro derivatives of 44, we have demonstrated that our bridge biaryl ketone is a suitable model to test the importance of the antiperiplanar effect as proposed by Cieplak or Anh. In addition to providing greater inflexibility over the normally employed cyclohexanone model, the calculations revealed, unlike 2-substituted cyclohexanones, only one energy minimum. The $\alpha$-fluoro derivative of our bridge biaryl ketone exhibits an extremely large preference for the equatorial conformation. However, for the $\alpha$-methylthio derivative of 44, the axial diastereomer is favoured, even in the highly polar solvent acetonitrile. We conducted a conformational analysis of 2-methoxy- and 2-methylthiocyclohexanone in five solvents using 500 MHz $\sp1$H NMR spectroscopy. Both compounds show a strong solvent dependence with the $\alpha$-methylthio group existing mainly in its axial conformer (n$\rm \sb{a}$ = 0.94) while the $\alpha$-methoxy group shows only a marginal preference in the same solvent (n$\rm \sb{a}$ = 0.56). Employing these results and those known for the $\alpha$-halo cyclohexanones provides the following order of increasing axial preference: fluoro$methoxy$chloro$bromo$methylthio. On the basis of molecular mechanics calculations, it is concluded that this order may simply reflect the varying strength in the van der Waals interaction between the carbonyl group and the 2-equatorial substituent. Our competitive reduction rate study has demonstrated with the aid of log ($\rm k\sb{R}\sp{app}/k\sb{H})$ and ($\rm k\sb{R}\sp{ac}/k\sb{H})$ versus $\sigma\sb1$ plots that the $\pi$-facial reactivity is mainly governed by sigma-inductive effects and not app orbital-orbital interactions as put forward by Cieplak or Anh. The results indicate an early transition state for reduction by LAH, while reduction by SBH exhibits a much later transition state. In both instances, the reactivity increased as the electron withdrawing ability of the $\alpha$-substituent increased. In the reduction using triethylsilane, the opposite trend was found. Since the $\pi$-facial reactivity for reduction with LAH shows the least influence from inductive effects, it should be the most likely to reveal an app effect. However, both the Cieplak and Anh theories do not predict the observed results. In addition, assuming that the face reactivity for the axial substituents represents a combination of inductive and app effects (Cieplak or Anh) and for the equatorial substituents represents only inductive effects, the study indicates, for all three reducing agents, a series of inconsistent results. The plots of log ($\rm k\sb{R}\sp{app}/k\sb{H}$) and ($\rm k\sb{R}\sp{ac}/k\sb{H}$) versus $\sigma\sb1$ do not exhibit a strong correlation. For all three reducing agents, the data for the methoxy substituents show the largest deviation. In addition to steric and torsional strain, we have reasoned this lack of linearity arises from a variable dipole-dipole interaction. (Abstract shortened by UMI.)