Thermal decomposition mechanisms of tert-alkyl peroxypivalates studied by the nitroxide radical trapping technique

Article abstract of DOI:10.1021/jo990680s

The thermolysis of a series of tert-alkyl peroxypivalates 1 in cumene has been investigated by using the nitroxide radical-trapping technique. tert-Alkoxyl radicals generated from the thermolysis underwent the unimolecular reactions, β-scission, and 1,5-H shift, competing with hydrogen abstraction from cumene. The absolute rate constants for β-scission of tert- alkoxyl radicals, which vary over 4 orders of magnitude, indicate the vastly different behavior of alkoxyl radicals. However, the radical generation efficiencies of 1 varied only slightly, from 53 (R = Me) to 63% (R = Bu(t)), supporting a mechanism involving concerted two-bond scission within the solvent cage to generate the tert-butyl radical, CO₂, and an alkoxyl radical. The thermolysis rate constants of tert-alkyl peroxypivalates 1 were influenced by both inductive and steric effects [Taft-Ingold equation, log(rel k(d)) = (0.97 ± 0.14)Σσ* - (0.31 ± 0.04)ΣE(s)(c), was obtained].

Full text of DOI:10.1021/jo990680s

16
J . Org. Chem. 2000, 65, 16-23
Th er m a l Decom p osition Mech a n ism s of ter t-Alk yl P er oxyp iva la tes
Stu d ied by th e Nitr oxid e Ra d ica l Tr a p p in g Tech n iqu e
Tomoyuki Nakamura,*,† W. Ken Busfield,‡ Ian D. J enkins,‡ Ezio Rizzardo,§
San H. Thang,§ and Shuji Suyama†
Fine Chemicals and Polymers Research Laboratory, NOF Corporation, 82 Nishimon,
Taketoyo-cho, Chita-gun, Aichi 470-2398, J apan, Faculty of Science, Griffith University,
Nathan, Queensland 4111, Australia, and CSIRO Molecular Science, Bag 10, Clayton South,
Victoria 3169, Australia
Received April 22, 1999
The thermolysis of a series of tert-alkyl peroxypivalates 1 in cumene has been investigated by using
the nitroxide radical-trapping technique. tert-Alkoxyl radicals generated from the thermolysis
underwent the unimolecular reactions, â-scission, and 1,5-H shift, competing with hydrogen
abstraction from cumene. The absolute rate constants for â-scission of tert-alkoxyl radicals, which
vary over 4 orders of magnitude, indicate the vastly different behavior of alkoxyl radicals. However,
the radical generation efficiencies of 1 varied only slightly, from 53 (R ) Me) to 63% (R ) But),
supporting a mechanism involving concerted two-bond scission within the solvent cage to generate
the tert-butyl radical, CO2, and an alkoxyl radical. The thermolysis rate constants of tert-alkyl
peroxypivalates 1 were influenced by both inductive and steric effects [Taft-Ingold equation, log-
c
(rel kd) ) (0.97 ( 0.14)Σσ* - (0.31 ( 0.04)ΣEs , was obtained].
In tr od u ction
on the product ratio of acetone to the corresponding
alcohol RCMe2OH without any consideration of the
production of the alcohol by other reactions.5 It was
concluded that the thermolysis rates of 1 were enhanced
by both hyperconjugative and inductive stabilizing effects
and that steric effects did not contribute to the rate
enhancement. Recently, Hendrickson et al. have re-
ported6 that both steric and electronic factors influence
the decomposition rate constants of t-alkyl tert-butyl
peroxides 5.
In this study, the thermolysis of a series of tert-alkyl
peroxypivalates 1a -g in cumene has been investigated
by using the radical trapping technique employing 1,1,3,3-
tetramethyl-2,3-dihydro-1H-isoindol-2-yloxyl (T), focusing
on the detailed behavior of tert-alkoxyl radicals and the
decomposition mechanism. The technique was developed
by the CSIRO and has been mostly used to investigate
initiation mechanisms in free radical polymerization.7
The aminoxyl T reacts with carbon-centered radicals (but
not oxygen-centered radicals) at close to diffusion-
Peroxyesters are among the most commonly used free
radical initiators for the radical polymerization of olefinic
compounds such as styrene and (meth)acrylates.1 They
are also useful radical sources for kinetic studies in
radical chemistry.2 tert-Alkyl peroxypivalates, such as
tert-butyl peroxypivalate 1a , are some of the most useful
peroxyesters in polymer manufacturing, and there have
been several reports concerning their decomposition
mechanisms.3,4 Peroxyesters 1 thermolyze homolitically
via a concerted two-bond scission with no cage return (k-1
) 0 in Scheme 1), as demonstrated by the quantitative
formation of carbon dioxide and the independence of the
decomposition rate on solvent viscosity.3a
The decomposition involves a proportion of cage reac-
tion, in which stable compounds are formed directly from
the transition state 2, e.g., isobutylene and tert-butyl
alcohol formed by disproportionation and di-tert-butyl
ether by radical recombination (eqs 2 and 3 in Scheme
1) for peroxyester 1a .3b However, there has been one
report suggesting a transition state 4 involving a con-
certed “three-bond” scission for the thermolysis of 1
(Scheme 2).4 This decomposition mechanism was based
* To whom correspondence should be addressed. Phone: +81 569
72 1403. Fax: +81 569 74 0009. E-mail: tomoyuki_nakamura@nof.co.jp.
† NOF Corporation.
‡ Griffith University.
§ CSIRO Molecular Science.
(1) Sheppard, C. S. In Encyclopedia of Polymer Science and Engi-
neering, 2nd ed.; Klingsberg, A., Piccininni, R. M., Salvatore, A.,
Baldwin, T., Eds.; Wiley-Interscience: New York, 1988; Vol. 11, pp
1-21.
(2) For example, see: Busfield, W. K.; J enkins, I. D.; Monteiro, M.
J . Polymer 1997, 38, 165-171.
(3) (a) Bartlett, P. D.; Simons, D. M. J . Am. Chem. Soc. 1960, 82,
1753-1756. (b) Koenig, T.; Wolf, R. J . Am. Chem. Soc. 1967, 89, 2948-
2952. (c) Lorand, J . P.; Chodroff, S. D.; Wallace, R. W. J . Am. Chem.
Soc. 1968, 90, 5266-5267. (d) Pryor, W. A.; Morkved, E. H.; Bickley,
H. T. J . Org. Chem. 1972, 37, 1999-2005. (e) Ernst, J . E.; Thankachan,
C.; Tidwell, T. T. J . Org. Chem. 1974, 39, 3614-3615.
(4) Komai, T.; Matsuyama, K.; Matsushima, M. Bull. Chem. Soc.
J pn. 1988, 61, 1641-1646.
(5) The production of the tert-alcohol via intramolecular hydrogen
abstraction by tert-alkoxyl radicals also has to be considered.
(6) Hendrickson, W. H.; Nguyen, C. C.; Nguyen, J . T.; Simons, K.
T. Tetrahedron Lett. 1995, 36, 7217-7220.
(7) Busfield, W. K.; J enkins, I. D.; Nakamura, T.; Monteiro, M. J .;
Rizzardo, E.; Suyama, S.; Thang, S. H.; Le, P. V.; Zayas-Holdsworth,
C. I. Polym. Adv. Tech. 1998, 9, 94-100 and references contained
therein.
10.1021/jo990680s CCC: $19.00 © 2000 American Chemical Society
Published on Web 12/16/1999

Thermal Decomposition Mechanisms of tert-Alkyl Peroxypivalates
J . Org. Chem., Vol. 65, No. 1, 2000 17
Sch em e 1
Sch em e 2
controlled rates to produce stable alkoxyamines.8 In our
previous work,9 we have shown that the thermolysis of
tert-alkyl peroxypivalates is not affected by the presence
of T and equimolar amounts of tert-butyl and tert-alkoxyl
radicals 3 are generated.
of tert-alkoxyl radicals. Alkoxyamines 7 and 8 were
derived from hydrogen abstraction from cumene by tert-
alkoxyl radicals (it is unlikely that any of the alkyl
radicals are involved in significant H-abstraction from
cumene under these reaction conditions, since the rate
of trapping is extremely fast). No other cumene-derived
products were detected. Phenylalkyl radicals such as
Resu lts a n d Discu ssion
2-phenylpropyl radicals could rearrange via 1,2-phenyl
Th er m a l Decom p osition P r od u cts of 1 in th e
P r esen ce of T. tert-Alkyl peroxypivalates 1a -g were
prepared by the reaction of pivaloyl chloride with the
corresponding tert-alkyl hydroperoxides in alkaline solu-
tion according to the literature procedure.4,9 The thermal
decomposition of 1 (0.040 M) in cumene as solvent in the
presence of T (0.040 M) was carried out in vacuo at 60
°C for 3.0 h.10 Alkoxyamines formed by trapping of the
carbon-centered radicals derived from the peroxide or of
the radicals resulting from the reactions of the primary
radicals with cumene were analyzed by HPLC, HPLC-
MS, and NMR.
11
migration; however, this reaction is too slow (<104 s-1
)
to compete with trapping under the reaction conditions.
Alkoxyamines 9 and 10 were derived from the alkyl
radicals formed by â-scission and by rearrangement of
tert-alkoxyl radicals, respectively. Unfortunately, an
insufficient amount of 10f was available for complete
characterization by NMR. The tentative structure of 10f
1
is proposed on the basis of the MS and H NMR. A tiny
amount of an isomer of 10f was also detected. The
relative yields of tert-alkoxyl radical-derived products
7-10 to the yield of tert-butoxyamine 6 (taken as 100%)
are shown in Table 1, and a postulated reaction mecha-
nism for the decomposition of tert-alkoxyl radicals is
shown in Scheme 3.
It can be seen from Table 1 that the total yields of
alkoxyamines 7-10 were entirely comparable to the yield
of 6 in all runs, showing that the thermolysis of 1
generates an equimolar amount of tert-butyl and tert-
alkoxyl radicals.9 A slightly lower yield of alkoxyamine
(9) (a) Nakamura, T.; Busfield, W. K.; J enkins, I. D.; Rizzardo, E.;
Thang, S. H.; Suyama, S. J . Am. Chem. Soc. 1996, 118, 10824-10828.
(b) Nakamura, T.; Busfield, W. K.; J enkins, I. D.; Rizzardo, E.; Thang,
S. H.; Suyama, S. Macromolecules 1997, 30, 2843-2847. (c) Nakamura,
T.; Busfield, W. K.; J enkins, I. D.; Rizzardo, E.; Thang, S. H.; Suyama,
S. J . Org. Chem. 1997, 62, 5578?5582. (d) Nakamura, T.; Busfield, W.
K.; J enkins, I. D.; Rizzardo, E.; Thang, S. H.; Suyama, S. J . Am. Chem.
Soc. 1997, 119, 10987-10991. (e) Nakamura, T.; Watanabe, Y.; Tezuka,
H.; Busfield, W. K.; J enkins, I. D.; Rizzardo, E.; Thang, S. H.; Suyama,
S. Chem. Lett. 1997, 1093-1094. (f) Nakamura, T.; Busfield, W. K.;
J enkins, I. D.; Rizzardo, E.; Thang, S. H.; Suyama, S. Chem. Lett. 1998,
965-966. (g) Nakamura, T.; Busfield, W. K.; J enkins, I. D.; Rizzardo,
E.; Thang, S. H.; Suyama, S. Polymer 1999, 40, 1395-1401.
tert-Butyl radicals generated from the thermolysis of
1 were immediately trapped by T to form tert-bu-
toxyamine 6 [kT(But.) ) ca. 9 ×108 M-1 s-1].8 On the other
hand, a variety of products were formed from the reaction
(10) A relatively low concentration of T was used in order to try to
study the (competitive) reaction of alkyl radicals such as 1,2-phenyl
migration. However, under the condition of the reaction, T is still
present in excess because of the low conversion (<40%) and <65%
efficiency of generation of radicals from 1.
(11) The absolute rate constant for the neophyl rearrangement has
been estimated to be 6 × 103 s-1 at 60 °C (Lindsay, D. A.; Lusztyk, J .;
Ingold, K. U. J . Am. Chem. Soc. 1984, 106, 7078-7093).
(8) Beckwith, A. L. J .; Bowry, V. W.; Moad, G. J . Org. Chem. 1988,
53, 1632-1641.

18 J . Org. Chem., Vol. 65, No. 1, 2000
Nakamura et al.
Sch em e 3
Ta ble 1. Rela tive Yield s of ter t-Alk oxyl Ra d ica l-Der ived
P r od u cts in th e Th er m olysis of 1 in Cu m en e in th e
P r esen ce of T a t 60 ˚Ca
Ta ble 2. Rela tive Ra tes of 1,5-Hyd r ogen Sh ift ver su s
â-Scission of ter t-Alk oxyl Ra d ica ls (3c, 3d , a n d 3f) in
Cu m en e a t 60 °C
relative product yields (%)
δ-hydrogen
tert-alkoxyl
run radicals (3)
R in 3
Me
7
8
9a 9b-g
10
tert-alkoxyl radical (3)
R in 3
Prn
Me3CCH2
c-C6H11
no.
type
k
1.5H/kâ(R)
1
2b
3
4
5
6
7
8
3a
3a
3b
3c
3d
3e
3f
94.0
92.8
38.8
22.4
2.7
4.3
4.3
1.6
0.8
0.1
0.2
1.7
1.7
0.9 58.7
0.6 32.8
0.7 43.1
0.5 94.5
3c
3d
3f
3
9
2
primary
primary
secondary
1.3
1.2
0.03
Me
Et
Prn
43.3
53.5
Me3CCH2
with the literature value of 6.3 M-1 at 60 °C in benzene.14
Similarly, the individual reactivities of the R- and
â-hydrogens of cumene toward abstraction by 3a were
determined. The values, corrected for the statistical
factor, were kR-H/kâ(Me) ) 7.7 and kâ-H/kâ(Me) ) 0.060
M-1 (per C-H bond), respectively, and the relative
reactivity of â-H to R-H of cumene was kâ-H/kR-H ) 0.008
(per C-H).15 Thus, the reactivity of the â-H of cumene
seems to be very comparable to that of tert-butylbenzene
[kâ-H/kâ(Me) ) 0.050 M-1 at 60 °C (per C-H bond)].19
â-Scission of ter t-Alk oxyl Ra d ica ls. In contrast to
the tert-butoxyl radical, other tert-alkoxyl radicals 3b-g
underwent significant â-scission to form alkoxyamines
9b-g as the major reaction pathway (eq 8 in Scheme 3).
For 3c and 3d other pathways are equally important (i.e.,
1,5-hydrogen shift) as shown in Table 2. The relative
rates of â-scission to hydrogen abstraction from cumene,
kâ(R)/kH, for tert-alkoxyl radicals 3 were obtained from
the ratio of 9/(7 + 8). Assuming that kH for any tert-
alkoxyl radical has approximately the same value,20 the
absolute rate constants for â-scission of tert-alkoxyl
Pri
4.8
c-C6H11
But
0.6 trace 0.1 96.8
0.2 trace 0.1 99.7
2.5c
3g
a
[1]0 ) 0.040 M, [T]0 ) 0.040 M, reaction time: 3.0 h. b Reaction
time: 65 h. c The total yield of two isomers (2.3% + 0.2%).
7 was observed in a longer reaction time (run 2 in Table
1), which might be due to lower stability of 7 under the
reaction conditions.12 The major reaction with tert-butoxyl
radicals 3a was hydrogen abstraction from cumene.
â-Scission to form methyl radicals was a very minor
process. On the other hand, tert-alkoxyl radicals 3b-g
underwent significant â-scission, generating alkyl radi-
cals to form the corresponding alkoxyamines 9b-g. In
addition, alkoxyl radicals 3c, 3d , and 3f underwent
intermolecular hydrogen abstraction (1,5-H shift) to form
10c, 10d , and 10f, respectively.
Hyd r ogen Abstr a ction fr om Cu m en e by ter t-
Alk oxyl Ra d ica ls. Cumene underwent hydrogen ab-
straction by tert-alkoxyl radicals, resulting in alkoxyamines
7 and 8 (eqs 5 and 6 in Scheme 3). The overall reactivity
of cumene toward hydrogen abstraction by tert-alkoxyl
radicals was estimated by using the competing fragmen-
tation reaction to form methyl radicals (eq 7) as a “free
radical clock”13 From the ratio of the product yield (7 +
8)/9a in run 1 and taking the concentration of cumene
as 7.2 M, the value of kH/kâ(Me) for tert-butoxyl radicals
3a was calculated to be 8.0 M-1, where kH is an overall
reaction rate constant for hydrogen abstraction from
cumene (per molecule). Our value is in good agreement
(14) Dulog, L.; David, K.-H. Makromol. Chem. 1976, 177, 1717-
1724.
(15) Although Walling and J acknow16 have obtained a much higher
value (kâ-H/kR-H ) 0.038) from the yields of 2-chloro-2-phenylpropane
11 (81.3%) and 1-chloro-2-phenyl-propane 12 (18.7%) in the photo-
chlorination of cumene by t-BuOCl at 40 °C, it was later shown that
the reaction involved a chlorine atom chain,17 which would lead to a
product distribution of 11:12 ) ca. 70:30.18
(16) Walling, C.; J acknow, B. B. J . Am. Chem. Soc. 1960, 82, 6108-
6112.
(17) Walling, C.; McGuinness, J . A. J . Am. Chem. Soc. 1969, 91,
2053-2058.
(18) Russell, G. A.; Brown, H. C. J . Am. Chem. Soc. 1955, 77, 4578-
4582.
(12) The half-life of the analogue alkoxyamine to 7, 1-(2-methyl-2-
phenylethoxy)-2,2,6,6-tetramethyl-4-oxopiperidine, has been reported
to be ca. 3.6 h at 100 °C in degassed tert-butylbenzene (Howard, J . A.;
Tait, J . C. J . Org. Chem. 1978, 43, 4279-4283).
(13) Griller, D. G.; Ingold, K. U. Acc. Chem. Res. 1980, 13, 317-
323.
(19) The value was calculated from the Arrhenius equation [log [A(â-
H abstraction)/A(â-scission)] ) -4.22 M-1, Ea(â-H abstraction) - Ea-
(â-scission) ) -24.7 kJ mol-1]. Howard, J . A.; Scaiano, J . C. Landolt-
Bo¨rnstein, New Series, Radical Reaction Rates in Solution; Fischer,
H., Ed.; Springer-Verlag: Berlin, 1984; Vol. 13, Part d, pp 107-108.
(20) Walling, C.; Padwa, A. J . Am. Chem. Soc. 1963, 85, 1593-1597.

Thermal Decomposition Mechanisms of tert-Alkyl Peroxypivalates
J . Org. Chem., Vol. 65, No. 1, 2000 19
radicals, kâ(R), were estimated from the concentration of
cumene (7.2 M) and the reported value of kH for 3a ()
8.7 × 105 M-1 s-1).21 The resulting values of kâ(R)
increased in the following order: 3a (1.1 × 105 s-1) < 3c
(8.9 × 106 s-1), 3b (9.1 × 106 s-1) < 3d (9.6 × 107 s-1) <
3e (1.2 × 108 s-1) < 3f (1.0 × 109 s-1) < 3g (3.1 × 109 s-1
)
It is obvious that the rate constants depend greatly on
the nature of the R group in tert-alkoxyl radicals, varying
by 4 orders of magnitude over the series 3a -g. Kochi has
mentioned that the rates for â-scission depend on the
stability of the leaving alkyl radicals.22 Our value of kâ
for 3a was very similar to the literature value of 0.7 ×
105 s-1 at 60 °C.23 However, it seems unlikely that the
rate constant for neopentyl radical elimination from 3d
is significantly higher (more than 10 times) than that of
n-propyl radical elimination from 3c and that of ethyl
radical elimination from 3b. The elimination rates to
produce neopentyl and n-propyl radicals in the â-scission
of the alkoxyl radical 13 have been reported to be the
same at 0 °C.24
F igu r e 1. 1,3-Diaxial interaction in the transition states for
1,5-H shift in tert-alkoxyl radicals 3c and 3d .
F igu r e 2. Axial (a) and equatorial (b) conformer of tert-alkoxyl
radical 3f.
Ta ble 3. Ra d ica l Gen er a tion Efficien cy (f) for ter t-Alk yl
P er oxyp iva la tes 1 in Cu m en e a t 60 °Ca
Me Et Prn Me3CCH2 Pri c-C6H11 But
peroxypivalates 1a 1b 1c
1d
1e
1f
1g
f (%)
53 53 54
61
55
56
63
This contradiction could be caused by the above as-
sumption of the same values of kH. We have previously
reported that the lower reactivity of alkoxyl radical 3d
is due to the steric hindrance around the oxygen radical
caused by the methyl groups on the γ-carbon as shown
in the preferred conformer 14 of 3d .9d
a
[1]0 ) 0.050 M, [T]0 ) 0.110 M.
â-scission, resulting in the corresponding alkoxyamines
10. The relative rates, k1,5H/kâ(R), for these tert-alkoxyl
radicals are listed in Table 2 as well as the numbers and
types of δ-position hydrogen.
Since the absolute rate constants of â-scission, kâ(R),
for both alkoxyl radicals 3c and 3d are comparable as
discussed above, the rate constants of k1,5H for both
alkoxyl radicals are also comparable, even though a 1,5-H
shift in 1,1,2,2-tetramethylpropoxyl radicals 3d has a
3-fold statistical advantage (3d has three times more
δ-hydrogens than tert-hexyloxyl radicals do). We have
previously suggested that the relatively low rate for the
1,5-H shift observed in 3d is a result of the steric
repulsion between the two methyl groups attached to the
R- and γ-carbons (see Figure 1),9d which may disfavor a
chair conformation of the six-membered ring as the
transition state for a 1,5-H shift.
Similarly, if there is significant steric hindrance around
the oxygen in other tert-alkoxyl radicals (although with
the exception of 3g, it is expected to be less than that in
3d ), the above kH values become smaller correspondingly.
It is also worth noting that the rate constants for
â-scission of 3a -g are less than the rate constant for
diffusion of a radical from a solvent cage (estimated to
be of the order of 1010 s-1).3d,25 Therefore, it is unlikely
that â-scission occurs via the transition state 4 in the
solvent cage as shown in Scheme 2. Further evidence
against the transition state 4 is provided by radical
efficiencies (see the following text).
The 1,5-H shift in 3f was a very minor reaction,
although 3f has secondary hydrogens at the δ-position.
This must be due to the large steric repulsion (1,3-diaxial
interactions) in the conformer (a) required for internal
H-abstraction (see Figure 2).
Ra d ica l Gen er a tion Efficien cy of th e Th er m olysis
of 1. The radical generation efficiency (f) of 1 was
measured by a trapping experiment. Thus, peroxyester
1 (0.050 M) in cumene was completely decomposed at 60
°C for 10 half-lives in the presence of excess T (0.110 M).
tert-Butyl radicals generated from the thermolysis of 1
were immediately trapped by T to form tert-butoxyamine
6 [kT(But.) ) ca. 9 × 108 M-1 s-1].8 In a separate
experiment, alkoxyamine 6 was found to be stable under
the reaction conditions (see the Experimental Section).
The value of f was determined from the yield of tert-
butoxyamine 6 on the basis of peroxyester consumed. The
results are shown in Table 3. Each value was the average
from several experiments. The observed value for 1a is
close to that given in previous reports, in which the extent
1,5-Hyd r ogen Sh ift of ter t-Alk oxyl Ra d ica ls. tert-
Alkoxyl radicals 3c, 3d , and 3f underwent the alternative
unimolecular reaction, a 1,5-H shift, in addition to
(21) Paul, H.; Small, R. D., J r.; Scaiano, J . C. J . Am. Chem. Soc.
1978, 100, 4520-4527.
(22) Kochi, J . K. J . Am. Chem. Soc. 1962, 84, 1193-1197.
(23) The value was calculated from the Arrhenius equation (log A
) 13.4 s-1, Ea ) 54.4 kJ mol-1) in ref 3d.
(24) Greene, F. D.; Savitz, M. L.; Osterholtz, F. D.; Lau, H. H.;
Smith, W. N.; Zanet, P. M. J . Org. Chem. 1963, 28, 55-64.
(25) Koenig, T.; Fischer, H. In Free Radicals; Kochi, J . K., Ed.; Wiley-
Interscience: New York, 1973; Vol. 1, p 173.

20 J . Org. Chem., Vol. 65, No. 1, 2000
Nakamura et al.
Ta ble 4. Decom p osition Ra te Con sta n ts a n d Activa tion P a r a m eter s for th e Th er m olysis of ter t-Alk yl P er oxyp iva la tes 1
in Cu m en ea
b
c d
1
R
T (°C)
kd × 105 (s-1
)
∆Hq (kJ mol-1
)
∆Sq (J mol-1 K-1
)
rel kd
Σσ* c
ΣEs
1a
1b
1c
1d
1e
1f
Me
Et
60
60
60
60
60
40
50
60
70
40
45
50
55
60
60
60
2.95
3.51
3.37
6.18
5.14
0.356
1.57
5.72
20.0
0.689
1.36
2.51
5.01
9.10e
6.91
10.32
118.3 ( 1.1
114.9 ( 1.2
116.3 ( 0.8
112.4 ( 1.3
114.3 ( 1.4
116.8 ( 0.7
22.7 ( 6.7
14.5 ( 7.2
18.0 ( 4.8
11.6 ( 8.1
15.8 ( 8.7
23.6 ( 4.0
t 1.00
1.19
1.14
2.09
1.74
1.94
-0.300
-0.315
-0.330
-0.344
-0.325
-0.260
-2.10
-2.35
-2.54
-3.43
-2.80
-3.01
Prn
Me3CCH2
Pri
c-C6H11
1g
But
109.4 ( 1.1
5.1 ( 7.0
3.08
-0.365
-3.70
1h
1i
PhCH2
Ph
116.5 ( 0.4
114.8 ( 0.2
24.4 ( 2.7
22.5 ( 1.3
2.34
3.50
-0.120
+0.025
-2.55
-2.94
a
b
[1]0 ) 0.05 M. See ref 4 for 1a -e,h and ref 30 for 1i. The relative thermolysis rate at 60 °C. c The sum of the substituent constant
of σ *.6,31,32 Hancock’s corrected steric constant32 calculated by eq 8 in ref 33. e Extrapolated.
d
of the cage reaction (100 - f)% for 1a in cumene was
estimated to be ca. 50%.3b,d
It is noteworthy that the efficiency does not change
very much by changing the alkyl group R in 1 despite
the very different behavior of these alkoxyl radicals.
Kiefer et al.26 and Leffler27 have studied radical efficiency
and mentioned that in a concerted decomposition, more
intervening molecular fragments between the caged
radicals cause higher radical generation efficiency be-
cause the reaction of caged radicals is hindered by such
molecules, thus allowing more of the radicals to escape.
Sch em e 4
F igu r e 3. ∆Sq plotted against ∆Hq for the thermolysis of tert-
alkyl peroxypivalates 1 in cumene.
99% radical efficiency.26,29 Thus, in the thermal decom-
position of peroxypivalates 1, the transition state 4
(which would result in two intervening molecules) can
probably be excluded; i.e., the extent of C-R bond
cleavage is not advanced in the transition state. The
small but significant increase of f for 1d and 1g compared
to the tert-butyl analogue 1a might be attributed to the
correspondingly longer distance between radical pairs
(tert-butyl and tert-alkoxyl radicals) caused by steric
repulsion, i.e., radical combination should be slightly less
for 1d and 1g.
Decom p osition Mech a n ism of ter t-Alk yl P er oxy-
p iva la tes in Cu m en e. The decomposition rate constants
and the activation parameters of the series of 1 [six
electron-donating alkyl substituents studied above, one
The value of f for di-tert-butyl monoperoxyoxalate 15,
weaker electron-donating (R ) PhCH2) and one electron-
withdrawing substituent (R ) Ph)] in cumene are sum-
marized in Table 4. It can be seen from Figure 3 that a
linear free energy relationship was observed in a plot of
∆Sq versus ∆Hq, which suggests that the fragmentation
mechanism does not change across the series.34 The
which decomposed via a concerted two-bond scission and
in which the caged radicals are separated by one CO2
molecule, was reported to be ca. 0.50.28 By contrast, the
decomposition of di-tert-butyl peroxyoxalate 16, which
forms two intervening molecules of CO2, resulted in 93-
(26) Kiefer, H.; Traylor, T. G. J . Am. Chem. Soc. 1967, 89, 6667-
6671.
(27) Leffler, J . E. An Introduction to Free Radicals; Wiley-Inter-
science: New York, 1993; pp 59-60.
(28) Bartlett, P. D.; Gontarev, B. A.; Sakurai, H. J . Am. Chem. Soc.
1962, 84, 3101-3104.
(29) (a) Bartlett, P. D.; Funahashi, T. J . Am. Chem. Soc. 1962, 84,
2596-2601. (b) Niki, E.; Kamiya, Y. J . Org. Chem. 1973, 38, 1403-
1406. (c) Niki, E.; Kamiya, Y. J . Am. Chem. Soc. 1974, 96, 2129-2133.
(30) Komai, T.; Kato, K.; Matsuyama, K. Bull. Chem. Soc. J pn. 1988,
61, 2641-2642.

Thermal Decomposition Mechanisms of tert-Alkyl Peroxypivalates
J . Org. Chem., Vol. 65, No. 1, 2000 21
F igu r e 5. Antiperiplanar conformer of 1.
F igu r e 4. Taft plot for the thermolysis of tert-alkyl peroxy-
pivalates 1 in cumene.
thermolysis rates of peroxypivalates 1 increased in the
order 1a < 1c < 1b < 1e < 1f < 1d < 1h < 1g < 1i. The
fact that peroxyester 1i decomposes 3.5 times faster than
1a can be explained by a polarized transition state 2′.30,35
However, it is clear from Figure 4 that the Taft equation
F igu r e 6. Correlation between relative rate constants for the
c
thermolysis of 1 and Hancock corrected steric constants (ΣEs ).
does not fully explain the rate acceleration of all the
peroxides. A small deviation of 1h (R ) PhCH2) from a
linear line passing through the points of 1a and 1i
indicates that the resonance effects on the alkyl group
stabilities are not important, which can be understood
by little possibility of the transition states 4 as mentioned
above.
It is worthy of note that the values of ∆Hq and ∆Sq for
two peroxypivalates with bulky alkyl groups [R ) Me3-
CCH2 (1d ) and But (1g)] are lower than those of others.
The trend in ∆Hq and ∆Sq has been explained in terms
of frozen bond rotations in the transition state.36 That
is, in the concerted two-bond scission of peroxyesters,
both of the σ bonds to the CO2 moiety need to be
approximately antiperiplanar, which results in the loss
of rotational freedom in the carbonyl to oxygen bond and,
hence, in a lower entropy (Figure 5).
c
F igu r e 7. Correlation of log(rel kd) with 0.97Σσ* - 0.31ΣEs .
It is very important to note that in the antiperiplanar
conformer the steric repulsion between oxygen in carbo-
nyl group and RCMe2 groups becomes significant due to
the reduced distance between carbonyl oxygen and
oxygen in alkoxyl groups and may cause the loss of
rotational freedom in an additional bond. We suggest that
the lower values of ∆Sq for peroxypivalates are a result
of the correspondingly restricted O-O bond rotation
caused by the steric hindrance around the oxygen in
alkoxyl groups as mentioned above. This may indicate
that the steric factors control the thermolysis rates of 1.
(31) Σσ * for R(CH3)2C group was obtained by adding each substitu-
ent constant of RCH2 and two CH3CH2. (a) Taft, R. W., J r. J . Am.
Chem. Soc. 1953, 75, 4231-4238. (b) Richardson, W. H.; Yelvington,
M. B. Andrist, A. H.; Ertley, E. W.; Smith, R. S.; J ohnson, T. D. J .
Org. Chem. 1973, 38, 4219-4225.
(32) Hancock, C. K.; Meyers, E. A.; Yager, B. J . J . Am. Chem. Soc.
1961, 83, 4211-4213.
(33) Fujita, T.; Takayama, C.; Nakajima. M. J . Org. Chem. 1973,
38, 1623-1630.
(34) Leffler, J . E. J . Org. Chem. 1955, 20, 1202-1231.
(35) (a) Bartlett, P. D.; Ru¨chardt, C. J . Am. Chem. Soc. 1960, 82,
1756-1762. (b) Koenig, T. In Free Radicals, Kochi, J . K., Ed.; Wiley-
Interscience: New York, 1973; Vol. 1, Chapter 3.
(36) (a) Bartlett P. D.; Hiatt, R. D. J . Am. Chem. Soc. 1958, 80,
1398-1405. (b) Leffler, J . E. An Introduction to Free Radicals; Wiley-
Interscience: New York, 1993; pp 139-140.
Hancock’s corrected steric constants (Es ),32 which
separate the hyperconjugation effect from Taft’s steric
substituent constant Es, were employed for the study of
the influence of steric effects on the thermolysis rate of
c
c
1. The values of Es were calculated by a method
developed by Fujita.33 The plots of the logarithm of the
thermolysis constants of peroxypivalates with electron-
donating alkyl substituents 1a -1g versus Esc of R(CH3)2C

22 J . Org. Chem., Vol. 65, No. 1, 2000
Nakamura et al.
the solvent under vacuum, a product of 6.1 g was obtained.
The purity was determined by iodometric titration using
isopropyl alcohol and acetic acid as the solvent and saturated
sodium iodide as the source of iodide. The purity and yield of
the hydroperoxide were 98.8% and 74.0%, respectively. The
structure was consistent with its NMR: δH(CDCl3) 1.42, (d,
6H, 2 × CH3, J ) 7.0 Hz), 2.6-2.8 (m, 4H, 2 × CH2), 3.3-3.5
(m, 2H, 2 × CH), 7.2-7.4 (m, 10H, 2 × ArH); δC(CDCl3) 21.5
(CH3), 36.5 (CH), 38.6 (CH2), 126.7, 126.9 (C-2, C-4), 128.8 (C-
3), 144.8 (C-1), 167.9 (CdO). The half-lives of the analogous
peroxides, 3-phenylpropionyl peroxide and 3-methyl-3-phenyl-
butyryl peroxide, were reported to be ca. 28 h at 55.8 °C and
ca. 30 h at 55.0 °C, respectively.38
Gen er a l Meth od . The general procedure for carrying out
the radical trapping experiments and for the quantitative
analysis of the reaction products has been described previ-
ously.9 The products of the trapping experiments were in each
case isolated by preparative HPLC and characterized by
spectroscopic methods.
groups as shown in Figure 6 give a good correlation with
a slope of -0.29 (r ) 0.968). These results indicate that
steric effects are a major contributor to the rate accelera-
tion for 1. Utilization of the Taft-Ingold equation [log-
c
(rel kd) ) F*Σσ * + δΣEs ] gives F* ) 0.97 ( 0.14 and δ )
-0.31 ( 0.04 (r ) 0.974) for the thermolysis of peroxy-
esters 1 (Figure 7). The positive value of F* shows that
an electron-withdrawing substituent R in the tert-alkoxyl
group accelerates the decomposition. Thus, the thermoly-
sis rate of tert-alkyl peroxypivalates 1 are influenced by
both polar (i.e., electronic) and steric effects, although a
previous report has excluded any contribution from steric
effects to the rate enhancement.4
Con clu sion s
In this study, the reactions of tert-alkoxyl radicals and
the thermolysis mechanism of 1 were studied by the
nitroxide trapping technique. The absolute rate constants
for â-scission of tert-alkoxyl radicals, which vary over 4
orders of magnitude, indicate the vastly different behav-
ior of alkoxyl radicals. However, even 1,1,2,2-tetrameth-
ylpropoxyl radicals 3g are relatively stable in the solvent
cage, with a kâ less than the rate constant for diffusion
from the cage. The radical generation efficiencies of a
series of peroxyesters 1 varied slightly, from 53 (R ) Me)
to 63% (R ) But). These values are not sufficiently high
to support the notion that â-scission occurs in the solvent
cage. Moreover, for R groups such as R ) Me3CCH2 or
Prn, the major reaction observed is a 1,5-H shift, not
â-fragmentation. Thus, the data presented here support
a homolysis mechanism involving concerted two-bond
scission within a solvent cage (Scheme 1), not three-bond
scission (Scheme 2), as has been suggested previously.
This work has clearly shown that the thermolysis rates
of a series of tert-alkyl peroxypivalates 1 correlate with
Kin etic Exp er im en t. The thermolysis of 1 (0.05 M), which
was carried out in degassed cumene at several temperatures
(40-70 °C), was measured by monitoring its disappearance
by iodometric titration and was found to satisfy first-order
kinetics at all of the temperatures. The decomposition rate
constants and the activation parameters at 60 °C were
calculated from Arrhenius plots.
Tr a p p in g Exp er im en ts. In a typical reaction, a solution
of 1 (0.040 M) and T (0.040 M) in cumene was degassed by
three successive freezing-pump-thaw cycles to 10-4 mmHg.
The reaction vessel was then sealed under vacuum and heated
at 60 ( 0.1 °C for 3.0 h. The majority (ca. 90%) of excess
cumene was then removed under reduced pressure prior to
analysis by reversed-phase HPLC with methanol/water mix-
tures as the eluent.
Ra d ica l Gen er a tion Efficien cy. A solution of 1 (0.050 M)
and excess T (0.110 M) in cumene was degassed. The reaction
vessel was then sealed under vacuum and heated at 60 °C for
65 h (>10 half-lives of 1). The majority (ca. 90%) of excess
cumene was removed under reduced pressure, which was
followed by the analysis of the residue by HPLC. Several runs
were carried out for each peroxyester, and the value of f was
determined from the average yield of tert-butoxyamine 6 based
on peroxyester consumed. A blank experiment was carried out
in order to test the stability of 6. Thermolysis of 6 (0.050 M)
in cumene in the presence of T (0.050 M) indicated no
significant decomposition of 6 after 65 h at 60 °C.
P r od u cts a n d New Com p ou n d s. The HPLC-separated
products were identified by electrospray mass spectrometry.
Products 6, 7, 9, and 10 (except for 10f) were also identified
by co-chromatography with authentic samples.7,39-41
The new compound 8 was isolated by preparative HPLC and
characterized by NMR. An authentic sample of alkoxyamine
8 was prepared by the thermolysis of di(3-phenylbutyryl)
peroxide in the presence of T in benzene. Thus, a solution of
di(3-phenylbutyryl) peroxide (97.9 mg, 0.30 mmol) and T (95.1
mg, 0.50 mmol) in benzene (6 mL) was degassed and heated
at 60 °C for 89 h. After evaporation of the solvent under
vacuum, alkoxyamine 8 was isolated from the residue by
preparative HPLC.
c
F*Σσ * + δΣEs , which indicates that the thermolysis rates
are accelerated by an electron-withdrawing substituent
in the tert-alkoxyl group, i.e., both inductive effects and
steric effects are important in the thermolysis reaction.
Among the tert-alkyl peroxypivalates 1 studied here, 1d
(R ) Me3CCH2) and 1g (R ) But) showed the highest
decomposition rates and radical generation efficiencies,
mainly as a result of steric effects.
Exp er im en ta l Section
Ma ter ia ls. Cumene was washed with concentrated H2SO4
and water, dried over anhydrous Na2SO4, distilled at atmo-
spheric pressure, and stored in a refrigerator (-20 °C). tert-
Alkyl peroxypivalates 14,9 and nitroxide T37 were prepared by
the literature procedure. 3-Phenylbutyryl chloride was ob-
tained from the chlorination of 3-phenylbutyric acid by thionyl
chloride. Spectroscopic data of 3-phenylbutyryl chloride: δH-
(CDCl3) 1.39, (d, 3H, CH3, J ) 6.9 Hz), 3.1-3.5 (m, 3H, CH
and CH2), 7.2-7.4 (m, 5H, ArH); δC(CDCl3) 21.4 (CH3), 36.7
(CH), 55.1 (CH2), 126.7, 127.0 (C-2, C-4), 128.9 (C-3), 144.0
(C-1), 172.4 (CdO).
P r ep a r a tion of Di(3-p h en ylbu tyr yl) P er oxid e. 3-Phen-
ylbutyryl chloride (9.13 g, 0.05 mol) in toluene (15 mL) was
added dropwise over a period of 25 min with stirring at 0-5
°C to a mixture of 30% hydrogen peroxide (3.40 g, 0.03 mol)
and 15% KOH (20.57 g, 0.055 mol). Stirring was continued
for 30 min at 0-5 °C, and then toluene was added to the
mixture. The organic layer was washed with water and dried
over anhydrous Na2SO4 and MgSO4. After the evaporation of
The tentative structure of 10f was proposed by means of
1
the MS and H NMR, and it is the most likely on the basis of
the other products observed in the reaction of other tert-alkoxyl
radicals 3c and 3d . However, the 1H NMR spectrum of
alkoxyamine 10f was complex, and an insufficient amount of
10f was available for complete characterization by NMR, which
usually requires proton decoupling experiments, DEPT, and
(38) Leffler, J . E.; Barbas, J . T. J . Am. Chem. Soc. 1981, 103, 7768-
7773.
(39) Rizzardo, E.; Serelis, A. K.; Solomon, D. H. Aust. J . Chem. 1982,
35, 2013-2024.
(40) Busfield, W. K.; Grice, I. D.; J enkins, I. D. Polym. Int. 1992,
27, 119-123.
(37) Griffith, P. G.; Moad, G.; Rizzardo, E.; Solomon, D. H. Aust. J .
Chem. 1983, 36, 397-401.
(41) Grant, R. D.; Griffith, P. G.; Moad, G.; Rizzardo, E.; Solomon,
D. H. Aust. J . Chem. 1983, 36, 2447-2454.

Thermal Decomposition Mechanisms of tert-Alkyl Peroxypivalates
J . Org. Chem., Vol. 65, No. 1, 2000 23
COSY relay experiments to confirm assignments. The four
strong methyl signals of the isoindoline moiety had chemical
shifts that overlapped those of the cyclohexyl ring protons, so
that determination of coupling constants and connectivity were
difficult. Another isomer of 10f was detected by HPLC and
HPLC-MS, but in lower yield (0.2%) than 10f (2.3%). It ran
slightly faster on HPLC (90% methanol/water mixture as
eluent).
Spectroscopic data for new compounds and some known
compounds are listed below [ring CH3 refers to methyl sub-
stituents on the isoindole and primed carbon numbers refer
to the monosubstituted phenyl ring (for 7 and 8) or the
cyclohexyl ring (for 9f)].
83.3 (CH2ON), 121.5 (C-4, C-7), 126.3 (C-2′, C-4′),127.2 (C-5,
C-6), 128.3 (C-3′), 144.9 (C-1′), 145.4 (C-3a, C-7a); m/z 332 (M
+ Na)+, 310 (M + H)+.
2-Cycloh exyloxy-1,1,3,3-t et r a m et h yl-2,3-d ih yd r o-1H -
isoin d ole 9f:41 δH(CDCl3) 1.1-2.2 (m, 22H, aliphatic H), 3.72
(m, 1H, CHON), 7.05-7.15 (m, 2H, ArH), 7.20-7.28 (m, 2H,
ArH); δC(CDCl3) 24.6 (C-3′), 25.3 and 30.5 (2 × br s, 4 × ring
CH3), 26.0 (C-4′), 32.6 (C-2′), 67.4 (C-1, C-3), 81.8 (CHON),
121.6 (C-4, C-7), 127.1 (C-5, C-6), 145.5 (C-3a, C-7a); m/z 296
(M + Na)+, 274 (M + H)+.
2-cis-[3-(1-H yd r oxy-1-m e t h yle t h yl)cycloh e xyloxy]-
1,1,3,3-tetr am eth yl-2,3-dih ydr o-1H-isoin dole 10f: δH(CDCl3)
1.24 and 1.26 [2 × s, 2 × 3H, (CH3)2COH], 1.35 and 1.51 (2 ×
br s, 2 × 6H, 4 × ring CH3), 0.8-2.4 [m, 10H, 4 × CH2 and
CH (cyclohexyl) and OH], 3.95-4.15 (m, 1H, CHON), 7.04-
7.15 (m, 2H, ArH), 7.18-7.30 (m, 2H, ArH); m/z 354 (M + Na)+,
332 (M + H)+.
2-(1-Meth yl-1-p h en yleth oxy)-1,1,3,3-tetr a m eth yl-2,3-d i-
h yd r o-1H-isoin d ole 7:9f δC(CD3OD) 26.3, 29.2 and 30.6 [4 ×
ring CH3 and 2 × (CH3)2CO)], 67.0 (C-1, C-3), 80.3 [(CH3)2CON],
122.6 (C-4, C-7), 127.4 (C-2′), 127.9 (C-4′), 128.2 (C-5, C-6),
128.8 (C-3′), 146.6 (C-3a, C-7a), 149.5 (C-1′).
2-(2-P h en ylp r op oxy)-1,1,3,3-tetr a m eth yl-2,3-d ih yd r o-
1H-isoin d ole 8: δH(CDCl3) 1.43, 1.44 and 1.48 (3 × br s, 15H,
4 × ring CH3 and CH3CH), 3.05-3.15 [m, 1H, CH3CH], 4.0-
4.2 [m, 2H, CH2ON], 7.10-7.15 (m, 2H, ArH), 7.22-7.32 (m,
2H, ArH), 7.34-7.40 (m, 5H ArH); δC(CDCl3) 18.7 (CH3CH),
24-31 (br hump, 4 × ring CH3), 39.8 (CH3CH), 67.3 (C-1, C-3),
Ack n ow led gm en t. We thank Ms. Catherine Rowen
for carrying out preliminary investigations. Financial
assistance from NOF Corp., Griffith University, and the
Australian Research Council is also acknowledged.
J O990680S