Thermal Stability of 2,2,6,6-Tetramethylpiperidine-1-oxyl
J. Am. Chem. Soc., Vol. 121, No. 27, 1999 6379
and 3:2 with respect to TEMPH. After 2 h at 423 K in AnH2
the product yield did not significantly change with time,
indicating that an equilibrium has been reached. By adding an
equimolar amount of benzoic acid, a quantitative conversion
of TEMPO into the amine occurred already within 1 h of
reaction time at 423 K. The acid was shown to be effective
only in the reduction of TEMPOH to TEMPH; addition of
benzoic acid to TEMPO in DPE did not change the inertness.
The stability of TEMPO was also studied in neat ethylbenzene
as a model for polystyrene. After 12 h at 423 K TEMPO was
converted for 38% into TEMPOH and TEMPH (see Table 3).
Taking into account that the polymerization usually lasts for
many hours, it is reasonable to assume that in the last stages of
the polymer growth, TEMPO is less effective. The yield of
styrene agreed with TEMPO reduction products as was found
with the other hydrogen-donor solvent. In another study it has
been demonstrated that heating TEMPO in ethylbenzene at 393
K for 12 h yielded the recombination product (not quantified)
between TEMPO and the 1-phenylethyl radical as was deter-
mined by NMR.27 Under our conditions, at 423 K, the
ethylbenzene-TEMPO adduct decomposes with a rate constant
of ca. 0.1 s-1, and even if the rate for disproportionation is
around 1% relative to recombination, it can be shown by kinetic
modeling that the final product after 12 h will be solely styrene.
A similar behavior has been observed for cumyl and TEMPO
at 353 K.22
Scheme 1
From the kinetic data, the carbon-oxygen bond dissociation
enthalpies at 298 K have been derived in the following way.
For the reverse reaction (see Scheme 1), the trapping of the
radical R• (cumyl, benzyl) with TEMPO, an apparent activation
enthalpy of about 1 kcal mol-1 has been determined,22,31 which
deviates from the temperature dependence found for near-
diffusion-limited reactions (3-4 kcal mol-1).
This phenomenon has been explained by considering that
upon reversible encounter (ka, k-a) of the two radical species, a
short-lived intermediate (C) is formed. Subsequent rearrange-
ment, through an activated process, renders R-TEMPO. At
sufficiently low temperatures the rate expression for product
formation is given by d[R-TEMPO]/dt ) kf[R•][TEMPO] with
kf ) kakb/(k-a + kb). Since ka is a known diffusion-controlled
rate constant, and kf has been determined experimentally, k-a
and kb can be obtained by kinetic simulation.32 On the basis of
the foregoing, and provided that the rate for scavenging of R•
and/or TEMPO is fast enough,33 the BDE(C-O) in the
N-alkoxyamine can be calculated from the measured activation
enthalpy for k-b by subtracting 1 kcal mol-1
.
Discussion
In Table 4, (recalculated) literature values for the BDE(C-
O) in N-alkoxyamines are presented, along with the results of
this work. Some data29,30 are not obtained under conditions of
kinetic isolation and therefore most likely are in error. Other
studies did recognize the presence of the persistent radical
effect.22,28
Thermal Stability of N-Alkoxyamines. This study demon-
strates the thermally labile nature of the carbon-oxygen bonds
in N-alkoxyamines (see Table 1). In fact, the combination of
TEMPO and cyclohexadienyl radical yields a compound with
a covalent bond of 18 kcal mol-1, and consequently the half-
life will be around 100 ms at room temperature. Although the
decomposition rate is high, the time frame of the photoacoustic
calorimetric method (less than 1 µs) prevents a convolution of
the heat deposition by the forward (eq 6) and the reverse
reaction.
On the basis of our data, excluding THF and triethylamine,
a plot of the BDE(C-O) vs the BDE(C-H) results in a linear
correlation of BDE(C-O) ) 1.04BDE(C-H) - 62.1 kcal mol-1
(see Figure 2) and spans a wide range of almost 30 kcal mol-1
.
The other recalculated data (see Table 4) nicely fit this
relation. According to the Pauling equation34 the BDE consists
of covalent and ionic contributions. From the slope of almost
one in the correlation we can conclude that the variation in the
electronegativity between the C-O and the C-H bonds remains
essentially constant throughout this series of compounds. This
correlation can be quite beneficial to predict the carbon-oxygen
bonds in various TEMPO-based N-alkoxyamines using the
known BDE(C-H) for the hydrocarbon. For example, in the
living polymerization of styrene, the carbon-oxygen bond is
The rates of thermal decomposition of some other N-
alkoxyamines have been determined on the basis of product
analysis by HPLC,28 NMR detection,29 or real time ESR.22,30
The relationship between Ea for unimolecular dissociation and
the BDE is rather straightforward for studies in the gas phase.
An additional enthalpic contribution may be anticipated when
reactions are studied in solution due to solvation of the reactant
and/or the transition state.23
(21) Hawari, J. A.; Engel, P. S.; Griller, D. Int. J. Chem. Kinet. 1985,
17, 1215-1219.
(22) Kothe, T.; Marque, S.; Martschke, R.; Popov, M.; Fischer, H. J.
Chem. Soc., Perkin Trans. 2 1998, 1553-1559.
(31) Chateauneuf, J.; Lusztyk, J.; Ingold, K. U. J. Org. Chem. 1988, 53,
1629-1632.
(23) The rate constant for homolysis of bibenzyl into two benzyl radicals
(32) For example, the termination rate constant for two cumyl radicals
is given by (4 × 1011) exp(-3.4/RT) M-1 s-1, while the kf for cumyl and
TEMPO is (2.5 × 108) exp(-0.88/RT) M-1 s-1. If one accepts A-a/Ab )
103, the entropy gain from the complex (C) into two radical species is always
much higher than a rearrangement inside the solvent cage; fitting yields
Ea,-a - Ea,b ) 2.7 kcal mol-1. Hence, with Ea,a of 3.4 kcal mol-1, the
enthalpy level for the TS of the rearrangment into R-TEMPO is 0.7 kcal
mol-1 above that for the two separated species R• and TEMPO. Hence, the
slowness of the TEMPO trapping reaction is due to an entropic effect.
Indeed, in the same solvent, kf decreases by a factor of 20 between 9,10-
dihydroanthracenyl and cyclohexadienyl while the overall reaction enthalpy
remains the same.3c
depends on the phase in which it has been derived: gas phase k s-1
)
/
1015.25 exp(-62.3/RT), tetralin k s-1 ) 1016. 6 exp(-66.8/RT), hence kgas
kliquid ) 4 (500 K) and 1.9 (600 K).24
(24) Stein, S. E.; Robaugh, D. A.; Alfieri, A. D.; Miller, R. E. J. Am.
Chem. Soc. 1982, 104, 6567-6570.
(25) Benson, S. W.; Shaw, R. In Organic Peroxides; Swern, D., Ed.;
Wiley-Interscience: New York, 1970.
(26) Upon standing, in the presence of (air) oxygen, both TEMPH and
TEMPOH were slowly reoxidized to TEMPO. This reverse reaction may
also be beneficial in living free radical polymerization.
(27) Connolly, T. J.; Scaiano, J. C. Tetrahedron Lett. 1997, 38, 1133-
1136.
(33) The kinetic expression for decomposition of the N-alkoxyamine
(28) Skene, W. G.; Belt, S. T.; Connolly, T. J.; Hahn, P.; Scaiano, J. C.
Macromolecules 1998, 31, 9103-9105.
(29) Li, I.; Howell, B. A.; Matyjaszewski, K.; Shigemoto, T.; Smith, P.
B.; Priddy, D. B. Macromolecules 1995, 28, 6692-6693.
(30) Veregin, R. P. N.; Georges, M. K.; Hamer, G. K.; Kazmaier, P. M.
Macromolecules 1995, 28, 4391-4398.
obeys d[R-TEMPO]/dt ) (k-ak-b/(k-a + kb))[R-TEMPO] - (kakb/(k-a +
kb))[R•][TEMPO]. When upon formation R• is scavanged, i.e., V10 is larger
than Va, d[R-TEMPO]/dt ) kd[R-TEMPO], with kd ) (k-ak-b/(k-a + kb).
Using the ratio for k-a/kb as derived in ref 32, kd ) k-b
.
(34) Pauling, L. The Nature of the Chemical Bond, 3rd ed.; Cornell
University Press: Ithaca, NY, 1960.