Publications : 2012

In a recent work (Org. Lett., 2012, 14, 358), we showed that the rate constants k(d) for the C-ON bond homolysis of chemically activated alkoxyamines were subject to solvent effects. We then investigated solvent effects on the non-activated alkoxyamine 1 ((diethyl(1-(tert-butyl(1-(pyridin-4-yl)ethoxy) amino)-2,2-dimethylpropyl) phosphonate) and its N+-O- oxide activated version 2, using 14 solvents exhibiting different solvent parameters – dipolar moments mu, dielectrical constants epsilon, cohesive pressures c, Reichardt solvent polarity constants E-T, viscosity eta, hydrogen bond donor and hydrogen bond acceptor constants alpha and beta, respectively, and nitrogen hyperfine coupling constants a(N). Weak solvent effects were observed both for 1 (4-5-fold from n-octane to 2,2,2-trifluoroethanol TFE) and for 2 (2-fold from n-octane to water) although k(d) increased 27-fold in n-octane and 19-fold in TFE from 1 to 2. It was shown that the C-ON bond homolysis rate constant k(d) increased with the a(N) values, meaning that the stabilization of the nitroxide was the main factor involved in the solvent effect. Approaches relying on the Koppel-Palm and Kalmet-Abboud-Taft relationships failed to describe the solvent effect for all diastereoisomers of 1 and 2. Nevertheless, the solvent polarity/polarizability (pi*) and hydrogen bond donor (alpha) properties are the main effects involved in the solvent effects at TS and on products.

Nitrogen hyperfine coupling constants (hcc) aN of nitroxides measured by electron paramagnetic resonance are used to probe the cybotactic effect of solvents. A few articles (Janzen et al., Can. J. Chem. in 1982 for aH beta and Deng et al., J. Fluorine Chem. in 2002, both for aH beta and aF beta) dealt with the cybotactic effect on the hcc of atoms (hydrogen H, fluorine F, and phosphorus P) in beta positions of nitroxides. They show either no significant change or an increase in a beta with aN. However, an early report by Il’Yasov et al. (Theor. Exp. Chem. 1976, 656) described a dramatic change in aP beta with the solvent and an anti-correlation of aP beta with aN. We decided to reinvestigate that work using the N-(2-methylpropyl)-N-(1-diethylphosphono-2,2-dimethylpropyl)-N-oxyl radical that carries both hydrogen and phosphorus atoms on the same carbon attached to the nitroxy moiety and that is highly persistent, in contrast to the nitroxide used by Il’Yasov. We observe the same trends as those observed by Il’Yasov, except that the solvent cybotactic effect on aN and aP beta is dramatically weaker in our case. We showed that aP beta is correlated to both the cohesive pressure c and the normalized Reichardt polar constant ETN, and that hydrogen-bonding donor solvents exhibit a specific behavior.

In a recent work (Org. Lett. 2012, 14, 358), we showed that the rate constant k(d) of the chemically activated C-ON bond homolysis of alkoxyamines was subject to solvent effects. However, we showed that solvent effects were weak for the nonactivated alkoxyamine 1 (diethyl (1-(tert-butyl(1-(pyridin-4-yl)ethoxy)amino)-2,2-dimethylpropyl)phosphon ate) and its N+-O- oxide activated version 3. On the other hand, the activated N-methylated version 2 of 1 experienced a strong solvent effect for a radical reaction, i.e., a 24-fold increase in kd from tert-butylbenzene (tBuPh) to 2,2,2-trifluoroethanol (TFE). Good correlations were observed with the normalized Reichardt solvent polarity constant E-T(N) and the nitrogen hyperfine coupling constant of the released nitroxide a(N,SGI), meaning that the stabilization of the nitroxide played an important role in C-ON bond homolysis in alkoxyamines. The Kalmet-Abboud-Taft relationships described successfully the solvent effect for each diastereoismer of 1 and 2, as for example with the minor diastereoisomer of 1 log(k(d)’/s(-1)) = -4.84 + 0.37 pi* + 0.21 alpha and that of 2 log(k(d)’/s(-1)) = -3.11 + 0.37 pi* + 0.47 alpha, with pi* being the polarity/polarizability and a the hydrogen bond donor (HBD) ability of the solvent. Surprisingly, the HBD effect is larger for 2 than for 1, whereas no extra lone pair is available in 2. This amazing effect was ascribed to the solvation of the counteranion, which is expected to be better solvated in a HBD solvent.