the COCl radical (cf., 13) competitively with the latter’s
scission to CO and Cl radical.2,6,18-20
Computational studies can help us decide among these
alternatives. B3LYP/6-31G(d) calculations21-23 afforded
energies and structures for most of the reaction channels
represented by 7-12. The reactions of carbene 1 in dodecane
or cyclohexane-d12 are best approximated by the computa-
tional results in vacuo. Here, SNi′ transition state (TS) 8,
Figure 1. B3LYP/6-31G(d)-computed structures:21 (A) TS 8 for
SNi′ rearrangement in a vacuum; (B) TS 7 for SNi formation of
unrearranged allyl chloride in simulated MeCN.
which affords 2-re, has ∆H‡ ) 9.25 kcal/mol. Structure
298
11 is a bifurcation point leading to 2-un in one direction
and 3-un (by a 1,2-shift) in the other direction. It has ∆H‡
298
the homolysis of carbene 1 may account for all four of the
principal products observed in hydrocarbon solvents; cf.,
Scheme 1. Here, radical pairs 10-d2 and 14-d2 are presumed
to form and recombine within the solvent cage.25
) 7.01 kcal/mol, so that the TS leading to 11 (which we
have not located), must have ∆H‡ > 7 kcal/mol. Similarly,
structure 12 is also a bifurcation point, giving either 2-un
or 3-re (SNi′). It features a high activation enthalpy (12.2
kcal/mol) and can be disregarded. The computed structure
for SNi′ TS 8 appears in Figure 1A. However, we were unable
to locate a TS for SNi fragmentation (7) in a vacuum, nor
could we find a minimum corresponding to ion pair 9 under
this condition.
Scheme 1
Most importantly, we calculate that homolysis of carbene
1 to radical pair 10 requires only ∼5.5 kcal/mol in a vacuum
and may well be the favored process in hydrocarbon
solution.24 Given that the further scission of •COCl into CO
and Cl• is expected to be rapid at 298 K2,6,18 and should be
competitive with allyl/COCl recombination, we suggest that
Experimentally, 2-un and 3-un are formed in greater
quantities than their rearranged counterparts 2-re and 3-re
in C6D12 (40:13 and 40:7, respectively, for thermolysis), so
that both radical pairs 10 and 14 appear to recombine
predominantly at the initial scission site of carbene 1. We
also note the large excess of 3-un over 3-re (40:7), which
could reflect the competition between recombination of allyl
and COCl radicals and the dissociation of the COCl radical
to CO and Cl. Alternatively, the energetically closest
competitor to carbene homolysis in vacuo is the bifurcation
leading to 2-un and 3-un. Participation of this channel could
also account for the dominance of the 2 unrearranged
products in C6D12.
In polar solvents, the situation is markedly altered.
Decomposition of 1 in MeCN (ꢀ ) 37.5) or DCE (ꢀ ) 10.36)
affords only the fragmentation products 2 and (in MeCN)
amide 6, while in CDCl3 (ꢀ ) 4.8 for CHCl3), about 9% of
3 is formed along with 83% of 2 (and ∼8% of the carbene
capture product, allyl dichloromethyl ether). Moreover,
labeling experiments in 9:1 DCE/CDCl3 reveal significant
allylic rearrangement in the formation of 2, with 2-re/2-un
) 34:66 (hν) or 41:59 (∆).
(16) 1H NMR analysis relied on the δ 3.90 d (J ) 6.6 Hz) signal of the
R-CH2 of 2-re, the δ 5.30 dd (J ) 1.5, 15.6 Hz) signal of the terminal
vinyl proton cis to CD2Cl of 2-un, the δ 3.48 d (J ) 6.6 Hz) signal of the
R-CH2 of 3-re, and the δ 5.10 dd (J ) 1.5, 10.2 Hz) signal of the terminal
vinyl proton trans to CD2COCl of 3-un. (Note that there are small solvent-
induced chemical shift changes between DCE-CDCl3 and C6D12-pentane
NMR solvents.)
(17) Laser flash photolysis experiments with diazirine 5 give kfrag ) (6.5
( 0.1) × 105 s-1 for carbene 1 in DCE and (6.6 ( 0.2) × 105 s-1 in MeCN.
These rate constants are not significantly affected by the addition of up to
1.8 M Bu4N+Cl-, indicating that neither SN2 nor SN2′ processes contribute
to the fragmentation of 1 in these solvents. Details appear in: Ma, Y. Ph.D.
Dissertation, Rutgers University, New Brunswick, NJ, 2003; pp 92f.
(18) ∆H‡
for the dissociation of COCl radical is ∼8 kcal/mol,19
298
whereas the COCl anion decays spontaneously to CO + Cl-.20 Therefore,
a COCl radical generated by homolysis of 1 might be recaptured by allyl
radical to yield 3-un or 3-re. However, a COCl anion from the heterolysis
of 1 would dissociate to CO and Cl- too rapidly for capture by allyl cation;
only 2-un and 2-re would be expected from the heterolysis of 1.
(19) Nicovitch, J. M.; Kreutter, P. H.; Wine, P. H. J. Chem. Phys. 1990,
92, 3539.
(20) Karpas, Z.; Klein, F. S. Int. J. Mass Spectrom. Ion Phys. 1976, 22,
189.
(21) All structures were fully optimized by analytical gradient methods
using the Gaussian 98 and Gaussian 03 suites22 and density functional (DFT)
calculations at the 6-31G(d) level, the exchange potentials of Becke,23a and
the correlation functional of Lee, Yang, and Parr.23b Activation energies
were corrected for zero-point energy differences (ZPVE) (unscaled) and
thermal effects at 298.150 K. Vibrational analyses established the nature
of all stationary points as either energy minima (no imaginary frequencies)
or first-order saddle points (one imaginary frequency).
(22) Gaussian 03, revision B.03; Gaussian, Inc: Pittsburgh, PA, 2003.
See Supporting Information for the full reference.
(23) (a) Becke, A. D. J. Chem. Phys. 1993, 98, 5648. (b) Lee, C.; Yang,
W.; Paar, R. G. Phys. ReV. B. 1988, 37, 785.
(24) Homolysis energy was calculated as the difference between the
computed energies of carbene 1 and the allyl plus COCl radicals.
(25) Homolysis of 1 to allyl and COCl radicals is analogous to the
cleavage of allylmethoxycarbenes to allyl and methoxycarbonyl radicals:
Venneri, P. C.; Warkentin, J. J. Am. Chem. Soc. 1998, 120, 11182. See
also the homolysis of benzyloxymethoxycarbene (Merkley, N.; El-Saidi,
M.; Warkentin, J. Can. J. Chem. 2000, 78, 356) and, more distantly, the
radical fragmentation of hydrazinoaminocarbenes: Cattoen, X.; Miqueu,
K.; Gornitzka, H.; Bourissou, D.; Bertrand, G. J. Am. Chem. Soc. 2005,
127, 3292.
Org. Lett., Vol. 7, No. 10, 2005
2057