3714 J. Am. Chem. Soc., Vol. 123, No. 16, 2001
Engel et al.
Although only 0.015% of cumyl radicals will add to styrene at
our styrene/TEMPO ratio of 150:1, the addition is reversible at
elevated temperatures, the essential feature of living free-radical
polymerization.
by analyzing their hydrolysis products, as was done so elegantly
in the earlier studies.
Although further experiments would be required to fully
establish the nature of the polymers initiated by 7, several facts
indicate that this trisazoalkane is a star polymer initiator: (1)
the formation of 1,3,5-triisopropylbenzene as the major ther-
molysis product with PhSH (cf. Table 9, Supporting Informa-
tion) and of 32a with ABNO show that 7 generates three cumyl-
type radicals on the same benzene ring, (2) the experiments
with 4, TEMPO, and styrene indicate that the cumyl radical is
the major initiator, (3) PMMA prepared from 7 contains an
aromatic ring, (4) a polymerization using 7, styrene, TEMPO,
and other additives yields a polymer of low polydispersity with
the expected GPC peak shape, and (5) three-arm ester-linked
initiators similar to 7 are known to generate star polymers.53,54
In summary, we have shown that the first deazatation rate of
tert-butylazocumenes 4-7 increases approximately statistically
with the number of pendant azo groups. However, many of the
“first-order” plots of nitrogen evolution are curved due to
deviations from this statistical ratio and to deazatation of various
secondary products such as 16. The rate constant of 5 is higher
than that of 6 and the product distribution of 5 is abnormal
because it decomposes via an azo-containing radical 34 to the
substituted quinonemethide 33. Although it is far more stable
than the parent quinonemethide, 33 polymerizes at room
temperature over the course of ∼20 h. tert-Butyl-cumyl radical
pairs disproportionate in both possible directions, but the cumyl
radical is a better H• donor than an acceptor. These radicals
also couple at the para position 27% as much as at the R
position. Cumyl radicals trapped by ABNO lead to 29, which
is a far more stable compound than the TEMPO adduct 31. The
system 4 plus TEMPO can act like a “one-radical” initiator
because TEMPO traps the tert-butyl radical, while the thermally
labile 31 is available to initiate the living free-radical poly-
merization of styrene. Finally, the weight of evidence indicates
that trisazoalkane 7 is an initiator of star free-radical polymer-
ization.
To determine whether TEMPO actually prevents tert-butyl
radicals from adding to styrene, we prepared a benzene solution
of 0.030 M 4, 4.0 M styrene, 0.0502 M TEMPO, and 0.0192
M undecane as internal standard. The degassed, sealed solution
was heated at 120 °C for 3 h, opened, and immediately analyzed
by GC. From the measured response factor of the GC stable
39, we determined its concentration to be 0.0141 M, corre-
sponding to 47% of the tert-butyl radicals. Of the remaining
53% of tert-butyl radicals, 40% were found as tert-heptyl-
benzene (10.5%) and as isobutane and isobutene (29.7%). This
experiment was repeated with twice the amount of TEMPO,
giving 39 (61%), tert-heptylbenzene (9.4%), and the C4 gases
(29.1%). The tert-butyl product balance of 87-100% shows
that only a minor fraction of these radicals are available to attack
styrene.
To be sure that 39 does not release tert-butyl radicals at 120
°C, we thermolyzed it separately in the presence of thiophenol,47
similar to our study of 29. A solution containing 39, PhSH (4
equiv), and (MeO)2CO (peak area standard) in C6D6 was heated
in a 120.0 °C bath. The reaction was monitored by 1H NMR of
both starting material and products, yielding first-order plots
with an average rate constant of 2.4 × 10-5 s-1. The formation
of large amounts of isobutene suggests a concerted Cope-type
elimination.50,79 Such a mechanism is consistent with the
unusually low activation energy for thermolysis of 39 (33.7 kcal/
mol from our observed rate constant and an assumed log A of
14.1) relative to the predicted value of 38.3 kcal/mol.51 Unlike
the concerted process, even minor amounts of tert-butyl radicals
from 39 could be problematic in the 4 + TEMPO initiator
system; however, homolysis of 39 can be precluded by replacing
tert-butyl with a bridgehead radical.
Preliminary experiments with 7 suggest that this trisazoalkane
can be used to prepare star polymers, which are of interest
because of their unique shape and their processing advan-
tages.53,55 A solution of 7 and methyl methacrylate in benzene
was heated to 122 °C for 20 min. The NMR spectrum of the
purified polymer showed a broad NMR singlet at 7.0 ppm, and
its UV spectrum was similar to that of 1,3,5-triisopropylbenzene.
This result shows that the aromatic ring of 7 is present in the
PMMA. In a second experiment, a styrene solution of 7,
TEMPO, benzoic acid,80 and 2-fluoro-1-methylpyridinium p-
toluenesulfonate (FMPTS)81 was degassed, sealed, and heated
at 125 °C for 73 h. The concentration of TEMPO was chosen
to react with all free radicals from 7, using the 16% cage effect
of 4 as a guideline. The isolated, reprecipitated polymer (69%
yield) exhibited a molecular weight Mn of 13 050 and a
polydispersity of 1.27. The GPC trace showed a major peak at
molecular weight 14 750 with a shoulder at MW ∼33 000, a
behavior characteristic of dimerization of growing star polymer
chains.82 Star polymers made from 7 contain no ester linkages
such as those employed in the previous cases.53,54 While this
structural feature is useful in making hydrolytically stable
polymers,83 it prevented us from characterizing the polymers
Experimental Section
General Methods. Anhydrous ether and THF were prepared by
distillation from benzophenone ketyl. All NMR spectra were obtained
in CDCl3 unless otherwise specified. The analytical instruments have
been described previously32 except for the Bruker Avance500 NMR
1
spectrometer used to acquire the low-temperature H and 13C NMR
spectra of 33. Gas chromatographic analyses employed an HP-5
capillary column: injector 200 °C, detector 250 °C, column 5 min at
50 °C, 10 °C/min to 250 °C, and hold 10 min. The mole/mole values
listed in Tables 4-7, 9, and 10 were calculated using the following
formula: (Ax/Astd)(Wstd)(1/MWx)(MWazo/Wazo), where A ) GC peak area,
W ) weight (mg), MW ) molecular weight, std ) internal standard,
x ) product, and azo ) azoalkane starting material.
Nitrogen Evolution Rate Measurements. A 2-mL flask containing
a weighed amount of azo compound (13, 7, and 4 mmol for 4, 5, and
7, respectively) and 1 mL of dodecane was connected to the gas
evolution kinetics apparatus.32 The solution was degassed, backfilled
with nitrogen, and immersed in a constant-temperature oil bath. The
decomposition was monitored on a chart recorder, whose readings were
taken manually and plugged into the computer program Kaleidagraph.
Cage Effect of 4 by the Excess Scavenger Technique. A benzene
solution of 0.094 M ABNO and 0.0282 M 4 was transferred into a
1-cm square tube with a 10/30 joint. After three cycles of freeze-
pump-thaw, the tube was sealed under vacuum and the absorbance
was determined to be 0.77 at 470 nm. The tube was then placed into
a constant-temperature bath at 110 °C for 2.5 h, where t1/2 of 4 is 28
min. After cooling to room temperature, the absorbance of ABNO was
found to be 0.39. Heating the tube at 110 °C for another 50 min led to
essentially no change in UV absorbance. The decrease of ABNO
(79) We gratefully acknowledge helpful discussion with Professor Hanns
Fischer, who has made similar observations.
(80) Georges, M. K.; Kee, R. A.; Veregin, R. P. N.; Hamer, G. K.;
Kazmaier, P. M. J. Phys. Org. Chem. 1995, 8, 301-305.
(81) Odell, P. G.; Veregin, R. P. N.; Michalak, L. M.; Brousmiche, D.;
Georges, M. K. Macromolecules 1995, 28, 3-8455.
(82) We thank Dr. Craig J. Hawker for pointing out the significance of
the GPC peak shape.
(83) Quirk, R. P.; Tsai, Y. Macromolecules 1998, 31, 8016-8025.