Thermolysis of acylazo O-trimethylsilyl cyanohydrins: Azoalkanes yielding captodatively substituted radicals

Article abstract of DOI:10.1021/jo981372h

A new synthetic method has been developed for preparing azoalkanes bearing geminal α-cyano and α-trimethylsiloxy groups. While the symmetrical compound 5 decomposes near room temperature to afford, almost entirely, the C-C dimers, the unsymmetrical azoalkane 3 requires heating to 75 °C. The thermolysis activation parameters of 3 and 5 can be rationalized on steric grounds without invoking captodative radical stabilization. A ¹³C NMR product study of photolyzed 3 in the presence of TEMPO at 21 °C shows that the fate of caged [tert-butyl, 1-trimethylsiloxy-1-cyanoethyl (14)] radical pairs is disproportionation, 17%, cage recombination, 20%, and cage escape, 63%.

Full text of DOI:10.1021/jo981372h

J . Org. Chem. 1998, 63, 9763-9768
9763
Th er m olysis of Acyla zo O-Tr im eth ylsilyl Cya n oh yd r in s:
Azoa lk a n es Yield in g Ca p tod a tively Su bstitu ted Ra d ica ls
Charles F. Billera, Travis B. Dunn, David A. Barry, and Paul S. Engel*
Department of Chemistry, Rice University, P.O. Box 1892, Houston, Texas 77251
Received J uly 14, 1998
A new synthetic method has been developed for preparing azoalkanes bearing geminal R-cyano
and R-trimethylsiloxy groups. While the symmetrical compound 5 decomposes near room temper-
ature to afford, almost entirely, the C-C dimers, the unsymmetrical azoalkane 3 requires heating
to 75 °C. The thermolysis activation parameters of 3 and 5 can be rationalized on steric grounds
without invoking captodative radical stabilization. A 13C NMR product study of photolyzed 3 in
the presence of TEMPO at 21 °C shows that the fate of caged [tert-butyl, 1-trimethylsiloxy-1-
cyanoethyl (14)] radical pairs is disproportionation, 17%, cage recombination, 20%, and cage escape,
63%.
The cross reactions of one reactive radical with another
can be studied only if both species are generated simul-
taneously. Of the possible precursors to a pair of unlike
radicals, azoalkanes are particularly convenient for
determining such parameters as the direction of dispro-
portionation (kd1/kd2) and the disproportionation-to-
recombination ratio (kd1 + kd2)/kc.1,2 Potential obstacles
to this approach include synthesis of the requisite azo-
alkane, accounting for self-reaction products of like
radicals, and secondary reactions of the initial radicals
with radicophilic olefin products.3,4 In some cases, azoal-
kanes have been used to estimate thermodynamic radical
stabilities5 since a plot of the activation energy for
symmetrical azoalkane thermolysis versus the bond
dissociation energy of the corresponding hydrocarbon is
a line of unit slope with correlation coefficient 0.99.6,7
synthetic approaches to azoalkanes,15-17 only a single
report18 has appeared previously on their use to generate
captodative radicals. Our product studies were carried
out with thiophenol as hydrogen donor and with the
persistent radical TEMPO to scavenge the initially
formed radicals and thus prevent secondary reactions.19
Resu lts
Azoalkanes 3 and 5 were synthesized from acyl diaz-
enes 220 and 421 by treatment with trimethylsilyl cyanide
(TMSCN).22,23 Symmetrical compound 5 was obtained as
a 50-50 mixture of meso and dl isomers, which were not
separated. While the unsymmetrical compound 3 was
stable at room temperature, 5 slowly decomposed. Nev-
ertheless, both azoalkanes could be purified by cold-
column chromatography on carefully dried silica gel.
Attempts to monitor the decomposition of 3 by UV
spectroscopy in toluene were foiled by the buildup of a
broad, unidentified, interfering absorption at 288 nm that
persisted even after heating for 15 h at 90 °C. Inclusion
of 1,4-cyclohexadiene or tert-butanethiol as a radical
(11) Viehe, H. G.; J anousek, Z.; Merenyi, R. Substituent Effects in
Radical Chemistry; D. Reidel: Dordrecht, 1986.
(12) Van Hoecke, M.; Borghese, A.; Penelle, J .; Merenyi, R.; Viehe,
H. G. Tetrahedron Lett. 1986, 27, 4569.
(13) Viehe, H. G.; Merenyi, R.; Stella, L.; J anousek, Z. Angew. Chem.,
Int. Ed. Engl. 1979, 18, 917.
(14) Viehe, H. G.; J anousek, Z.; Merenyi, R.; Stella, L. Acc. Chem.
Res. 1985, 18, 148.
(15) Timberlake, J . W.; Stowell, J . C. The Chemistry of the Hydrazo,
Azo, and Azoxy Groups; Wiley & Sons: New York, 1975.
(16) Lang-Fugmann, S. In Methoden Der Organichen Chemie;
Houben-Weyl, Ed.; G. Thieme: Stuttgart, 1992; Vol. E16d, Teil I, p
102.
Herein we report thermolysis kinetics and product
studies of two azoalkanes, prepared by a new method,
that decompose to captodative8-14 trimethylsilyl (TMS)
cyanohydrin radicals. Because of limitations in the
(17) Barton, D. H. R.; J aszberenyi, J . C.; Liu, W. S.; Shinada, T.
Tetrahedron 1996, 52, 14673.
(18) J anousek, Z.; Bougeois, J .-L.; Merenyi, R.; Viehe, H. G.;
Luedtke, A. E.; Gardner, S.; Timberlake, J . W. Tetrahedron Lett. 1988,
29, 3379.
(1) Neuman, R. C.; Aldaheff, E. S. J . Org. Chem. 1970, 35, 3401.
(2) Engel, P. S.; Wu, W.-X. J . Org. Chem. 1990, 55, 2720.
(3) Kelly, D. P.; Serelis, A. K.; Solomon, D. H.; Thompson, P. E. Aust.
J . Chem. 1987, 40, 1631.
(4) Hebeger, K.; Fischer, H. Int. J . Chem. Kinet. 1993, 25, 249.
(5) Luedtke, A.; Meng, K.; Timberlake, J . W. Tetrahedron Lett. 1987,
28, 4255.
(6) Engel, P. S. Chem. Rev. 1980, 80, 99-150.
(7) Al-Sader, B. H.; Crawford, R. J . Can. J . Chem. 1970, 48, 2745.
(8) Beckhaus, H.-D.; Ru¨chardt, C. Angew. Chem., Int. Ed. Engl.
1987, 26, 770.
(19) Nakamura, T.; Busfield, W. K.; J enkins, I. D.; Rizzardo, E.;
Thang, S. H.; Suyama, S. J . Am. Chem. Soc. 1997, 119, 10987.
(20) While N-acetyl N′-tert-butyl diazene (2) is a new compound, the
methyl and isopropyl analogues have been prepared. Lynch, T. R.;
MacLachlan, F. N.; Siu, Y. K. Can. J . Chem. 1971, 49, 1598.
(21) Campbell, J . A.; Mackay, D.; Sauer, T. D. Can. J . Chem. 1972,
50, 371.
(9) Korth, H.-G.; Sustmann, R.; Merenyi, R.; Viehe, H. G. J . Chem.
Soc., Perkin Trans. 2 1983, 67.
(10) Neumann, W. P.; Penenory, A.; Stewen, V.; Lehnig, M. J . Am.
Chem. Soc. 1989, 111, 5845.
(22) Rasmussen, J . K.; Heilmann, S. M.; Krepski, L. R. Advances
in Silicon Chemistry; J AI Press: Greenwich, CT, 1991; Vol. 1, p 65.
(23) Kobayaski, S.; Tsuchiya, Y.; Mukaiyama, T. Chem. Lett. 1991,
537.
10.1021/jo981372h CCC: $15.00 © 1998 American Chemical Society
Published on Web 12/04/1998

9764 J . Org. Chem., Vol. 63, No. 26, 1998
Billera et al.
eliminate NOE, was not noticeably broadened by the
presence of the paramagnetic compound. In fact, the
residual TEMPO was beneficial because it shortened T1
so that we were able to use a delay time of only 20 s
between pulses. Although the spectrum exhibited 46
peaks (cf. Table 3), an authentic sample of nearly every
product was available, allowing us to assign all but seven
peaks, or all but 3% of the total area. The chemical shift
of the unassigned peaks is far from that expected for a
ketenimine.29 The relative yield of each product was
calculated from the intensities of all of its 13C signals
combined, leading to the following results: isobutane, 8%;
6, 6%; isobutene, 3%; 7, 2%; 8, 11%; 15, 34%; 16, 35%;
and 17, 1%. Lending credence to these figures, we found
nearly equal amounts of each product expected from one
given pathway; e.g., cage escape afforded the same
quantity of 15 as 16.
scavenger did not eliminate the problem, and thiophenol
produced diphenyl disulfide, which interfered even more
with UV monitoring. Although 5 showed no 288-nm band
(see below), the kinetics of both azoalkanes were deter-
mined by nitrogen evolution.24
In the absence of a radical trap, thermolysis of 3
yielded a complex array of products, but inclusion of 0.5
M thiophenol as a scavenger led to a more tractable
mixture. The identified products and their GC-peak areas
relative to that of PhSSPh are as follows (compound, GC-
peak area): 6, 0.090; 7, 0.50; 8, 0.19; 9, 0.013; 10a , 0.023;
10b, 0.019; 11, 0.009; 12, 0.22; 13, 0.028; and PhSSPh,
1.00. Authentic samples of 6, 7, 8, 10a ,b, and 12 were
prepared independently, but the structures of minor
products 9, 11, and 13 are based only on GC/MS. Their
mode of formation is rationalized in Scheme 1 and in the
Discussion (see below).
Not surprisingly, the thermolysis products from 5
proved to be simpler than those from 3. When a solution
of 5 in C6D6 or CDCl3 was allowed to stand for 1 day, the
NMR spectrum showed, exclusively, the known32 dia-
stereomers 10a ,b and no olefin peaks. Thermolysis of 5
in the presence of excess TEMPO yielded 15, but little
of dimer 10a ,b, indicating that the cage effect is small.
This value was determined independently by the excess
scavenger technique33 with TEMPO to be only 15%. When
thiophenol was included as scavenger, one of the products
was 12, as proven by comparison with an authentic
sample.34 The major 1H NMR peaks showed 10a ,b, 12,
and 7 in the molar ratio 1.0:0.5:1.6, while the GC peak
area ratio was 1.0:0.3:1.0.
Since the 2-cyanopropyl radicals from AIBN25 and
other cyano substituted radicals26-28 produced a sizable
amount of ketenimine, a sample of 3 was thermolyzed
in toluene, the solvent was evaporated, and the residue
was dissolved in CCl4. The IR spectrum of this solution
showed only a small band at 2011 cm-1 25,26,29
indicating
,
either that the ketenimine hardly formed or that it
decomposed.
The IR spectrum of 5 in CCl4 at room temperature
exhibited a weak but sharp band at 2011 cm-1 that
remained unchanged for at least 24 h. Although this band
might be attributed to a small amount of ketenimine 18,
such a product would not be expected to persist.25,35 In a
A product study of 3, in the presence of TEMPO,
allowed us to calculate the disproportionation-to-recom-
bination ratio for the mixed radical pair, the direction of
disproportionation, and the cage effect. This clever ap-
proach was employed previously for a pair of radicals
differing only in their ester side chain.3 Because alkoxy-
amines are often unstable upon heating,30,31 we decom-
posed our new azoalkanes near ambient temperature and
employed quantitative 13C NMR as the analytical method.
A C6D6 solution of 3 and TEMPO in a sealed, degassed
NMR tube was irradiated at 366 nm until the color of
TEMPO had faded but not disappeared, at which point
some of 3 also remained. The 13C NMR spectrum of the
mixture, which was acquired using gated decoupling to
more determined effort to detect 18, a degassed, sealed
solution of 5 in hexane was heated at 40 °C with
monitoring by UV. The azo band at 368 nm disappeared
smoothly with no evidence of rising absorption at 290 nm,
the expected λmax of a ketenimine. As a control experi-
ment, a degassed, sealed solution of AIBN at 20 °C was
irradiated at 366 nm and monitored by UV. The azo band
at 344 nm disappeared while a new 290-nm absorption
maximum due to 19 grew in. In accord with the well-
(24) Engel, P. S.; Dalton, A. I.; Shen, L. J . Org. Chem. 1974, 39,
384.
(25) Hammond, G. S.; Wu, C.-H. S.; Trapp, O. D.; Warkentin, J .;
Keys, R. T. J . Am. Chem. Soc. 1960, 82, 5394.
(26) Zarkadis, A. K.; Neumann, W. P.; Dunnebacke, D.; Penenory,
A.; Stapel, R.; Stewen, U. Chem. Ber. 1993, 126, 1179.
(27) Neumann, W. P.; Stapel, R. Chem. Ber. 1986, 119, 3432.
(28) Lynch, T. R.; MacLachlan, F. N.; Suschitzky, J . L. Can. J . Chem.
1973, 51, 1378.
(32) Muller, R.; Neef, H. British Patent BG 680408, 1971; Chem.
Abstr. 1971, 75, 6078.
(29) Cunico, R. F.; Kuan, C. P. J . Org. Chem. 1992, 57, 1202.
(30) Engel, P. S.; Duan, S.; Arhancet, G. B. J . Org. Chem. 1997, 62,
3537.
(31) Grattan, D. W.; Carlsson, D. J .; Howard, J . A.; Wills, D. M.
Can. J . Chem. 1979, 57, 2834.
(33) Nelsen, S. F.; Bartlett, P. D. J . Am. Chem. Soc. 1966, 88, 143.
(34) Hertenstein, U.; Hunig, S.; Reichelt, H.; Schaller, R. Chem. Ber.
1986, 119, 699.
(35) Kim, S. S.; Liu, B.; Park, C. H.; Lee, K. H. J . Org. Chem. 1998,
63, 1571.

Use of Azoalkanes to Generate Captodative Radicals
J . Org. Chem., Vol. 63, No. 26, 1998 9765
Sch em e 1. Rea ction s of ter t-Bu tyl a n d MeC‚(CN)(OTMS) Ra d ica ls fr om Th er m olysis of 3 w ith P h SH a t 90
°C in C6D6
Ta ble 1. Activa tion P a r a m eter s of 3 a n d Rela ted
Un sym m etr ica l Azoa lk a n esa
Ta ble 2. Activa tion P a r a m eter s of 5 a n d Rela ted
Sym m etr ica l Azoa lk a n esa
∆Gq
(100 °C)
∆Gq
(100 °C)
entry t-Bu-NdN-R, R )
∆Hq
∆Sq
entry
R-NdN-R, R )
∆Hq
∆Sq
1
2
3
4
5
MeC(CN)(OTMS)b
Me2C(CN)
Me2CCHdCH2
Me3CCH2CMe2
31.3 ( 0.3 11.8 ( 0.9
26.9
28.6
26.8
33.5
26.0
1
2
3
4
5
6
MeC(CN)(OTMS)b
Me3C
Me2C(CN)
24.3 ( 1.2
42.1
30.1
40.9
25.9
3.9 ( 3.9
22.8
36.0
26.7
34.3
24.3
30.9
31.1
25.1c
34.0
26.8
38.1
14.7
-0.1
12.3
9.2
16.3
9.2
17.7
4.2
2.9
9.5
Me2C(OMe)
Me3CCH2C(CN)(Me) 29.4
Me2CCHdCH2
Me3CCH2CMe2
Me3CCH2CMe2
Me3CCH2C(CN)(Me)
32.0
34.6
a
b
Data from ref 6. This work.
7
known behavior of AIBN,25,36 the 290-nm band became
five times more intense than the original azo band. A
comparable solution of 5 was then irradiated at 0 °C,
causing only a small loss of azo absorption but no buildup
at 290 nm. Continued irradiation at 20 °C led to eventual
destruction of 5 but still no ketenimine band. If keten-
imine 18 is actually forming, it must be very labile
thermally or else its UV λmax is drastically blue shifted
from that of 19. Neither of these possibilities is likely.
a
b
Data from ref 6. This work. c ∆Gq (80 °C) calculated from a
single rate constant in ref 6.
combined effect of MeO (1.7 kcal/mol, entry 4) and CN
(9.3 kcal/mol, entry 3). In the symmetrical series, 5 is
1.5 kcal/mol more labile than the dimethylallyl analogue
(entry 5). This difference is in the opposite direction from
the 0.1 kcal/mol difference in entries 1 and 3 of Table 1,
but the low ∆Sq of entry 3 makes such a comparison
risky. Once again using a hydrocarbon model for steric
effects, we find that azo-tert-octane (entry 6) is 5.0 kcal/
mol more labile than ATB (entry 2), while the cyano
series (entries 3 and 7) shows a ∆Gq that is 1.6 kcal/mol
lower. Relief of steric compression during deazatation
suffices to explain the greater decrease in ∆Gq caused
by CN and OTMS acting together than the sum of the
changes in ∆Gq effected by them, taking MeO as an
electronic model for TMSO.37 Despite the fact that a plot
of azoalkane Ea versus hydrocarbon BDE is linear,6,7
experimental errors diminish the utility of Ea as a
measure of incipient radical stability. Relative thermoly-
sis rates or ∆Gq tends to be more reliable, but, as seen
even in the few cases in Tables 1 and 2, ∆Sq varies widely.
Another complication is that geminal substituents on
carbon can interact to change the ground-state en-
ergy,38,39 which is manifested as a polar effect on azoal-
kane thermolysis rates.40 While the data in Tables 1 and
Discu ssion
The thermolysis kinetics of 3 and 5 were cleanly first
order. The activation parameters for 3 are compared with
those of four related compounds in Table 1. Tentatively
using ∆Gq to evaluate the stability of incipient radicals,
we see that the combination of the OTMS and CN groups
is more stabilizing by 1.7 kcal/mol than a CN alone (entry
2). In fact, these two groups together are nearly as
effective as a methyl and vinyl moiety (entry 3). The 1.7
kcal/mol decrease in ∆Gq is probably of steric origin,
considering that tert-butylazo-tert-octane (entry 4) is 2.5
kcal/mol more labile than azo-tert-butane (ATB, ∆Gq (100
°C) ) 36.0 kcal/mol.) Similarly, replacement of methyl
by neopentyl in tert-butylazo-2-cyanopropane lowers ∆Gq
by 2.6 kcal/mol (cf. entries 2 and 5.)
Table 2 compares selected symmetrical azoalkanes
with 5. The lowering of ∆Gq by CN and OTMS, relative
to ATB, is 13.2 kcal/mol, which exceeds the 11 kcal/mol
(37) Creary, X. Unpublished work, 1998.
(38) Nicholas, A. M. d. P.; Arnold, D. R. Can. J . Chem. 1984, 62,
1850.
(36) Hammond, G. S.; Trapp, O. D.; Keys, R. T.; Neff, D. L. J . Am.
Chem. Soc. 1959, 81, 4878.
(39) Ruchardt, C. Angew. Chem., Int. Ed. Engl. 1970, 9, 830.

9766 J . Org. Chem., Vol. 63, No. 26, 1998
Billera et al.
2 and in ref 18 are not without interest, they cannot be
taken as evidence for or against the captodative effect.
The thermolysis products of 5 consisted solely of dimers
10a ,b, whose formation is in accord with the tendency
of delocalized radicals to recombine rather than dispro-
portionate.41 Similar behavior was seen in tert-butoxy-
(cyano)methyl radicals9 and in the radical formed by
addition of 2-cyanopropyl to silyl enol ether 6.34 Ther-
molysis of 5 with thiophenol led to sulfide 12, which
arises by cross recombination of phenylthiyl radicals and
14. A similar cross recombination product, the alkoxy-
amine, was formed upon thermolysis of azo-R-phenyl-
ethane with diethylhydroxylamine.30 An alternate source
of 12 is the known reaction34 of 6 with thiophenol, but
the absence of 6 without the scavenger precludes this
63% of the radicals escape the solvent cage, which is an
entirely reasonable figure by comparison to related
systems.43 A small amount of hydroxylamine 17 was
found among the products, suggesting the possibility that
TEMPO disproportionates with either tert-butyl radical
or with 14. A sample of 5 was allowed to decompose
thermally in the presence of TEMPO, and the product
mixture was analyzed by 13C NMR. Since no 17 was
detectable, this minor product from 3 must come either
from tert-butyl radicals plus TEMPO or from another
source altogether. Comparison of the present results with
the cumyl-cyclohexyl case, where recombination is also
the dominant self-reaction of the delocalized radical,
shows distinct similarities. Our total cage disproportion-
ation-to-recombination ratio (0.9) is lower than that of
cumyl-cyclohexyl (1.3) while the ratio of disproportion-
ation to the conjugated olefin 6 relative to the noncon-
jugated one (isobutene) is 3 in 3 versus 5 in the literature
example.1
In summary, we have developed a method to convert
acyldiazenes to hitherto inaccessible azoalkanes 3 and
5. The thermolysis rates of these azoalkanes do not
support captodative stabilization of TMS cyanohydrin
radical 14. The low cage effect for 5 (15%) is at odds with
our failure to detect ketenimine 18. J udging from the
observation that thiophenol does not completely prevent
3 from dimerizing to 10a ,b, we suggest that this hydro-
gen donor is only partially effective in trapping 14.
Finally, photolysis of 3 in the presence of excess TEMPO
followed by 13C NMR analysis allowed us to quantify the
competing processes in a pair of unlike radicals.
pathway. The relatively large amount of 10a ,b in the
presence of PhSH is only partly due to cage recombina-
tion; the rest is attributed to inefficient scavenging of 14
by PhSH, judging from the fact that 3 also affords 10a ,b
with PhSH (see below).
The low cage effect (15%) seen with 5 can be attributed
to steric hindrance to radical recombination9,42 but then
recombination should preferentially occur at cyano ni-
trogen. Not only did we find no evidence for ketenimine
18, but azo-tert-octane at 20 °C exhibited a normal cage
effect of 30%, despite the similar geometry of the tert-
octyl radical to 14. No appealing rationale for this
dilemma comes to mind. In contrast to 5, the cage effect
of the mixed radical pair t-Bu•/14 is also normal (37%).
The thermolysis products of 3 can be explained on the
basis of Scheme 1. In addition to the expected thiophenol-
scavenging product, 7, we find a sizable amount of
recombination product 8. The tert-butyl/14 radical pair
undergoes disproportionation in both directions, but the
resulting olefins isobutene and 6 suffer some attack by
other radicals to afford 9, 11, and 13. The addition of 14
to 6 to yield 11 is analogous to the reported addition of
2-cyanopropyl to 6,34 while the presence of 10a ,b shows
that 0.5 M thiophenol is not a completely effective
scavenger of 14. Unlike 5, the radicals from 3 dispropor-
tionate, in part, to give 6, which therefore cannot be ruled
out as the source of 12 for the unsymmetrical case.
Photolysis of 3 in the presence of TEMPO eliminates
the secondary reactions and allows us to assess the
relative importance of the three reaction pathways of the
caged tert-butyl/14 radical pair. Cage recombination to
afford 8 accounts for 20% of the radicals, as calculated
from %8/[%8 + 0.5(%isobutane + %6) + 0.5(%isobutene
+ %7) + 0.5(%15 + %16)], while cage disproportionation
accounts for 17%. Disproportionation to 6 and isobutane
is more important than that to 7 and isobutene, presum-
ably because 14 is more hindered than tert-butyl and is
therefore more sensitive to steric effects.3 The remaining
Exp er im en ta l Section
Gen er a l Meth od s. Most of these have been described
earlier.30 The NMR solvents CDCl3 and C6D6 from Cambridge
Isotope Laboratory were used without further purification. The
250 MHz NMR chemical shifts (δ, ppm) are stated relative to
internal TMS, hexamethyldisiloxane (1H δ ) 0.115), or solvent
signal (CDCl3 1H δ ) 7.26, 13C δ ) 77.0; C6D6 1H δ ) 7.15, 13
C
δ ) 128.0). Analytical GC was carried out with a DB-5
capillary column (0.25 mm × 30 m) and FID. The GC
conditions were as follows: head pressure, 16 psi; He split flow,
53 mL/min; injector, 210 °C; detector, 250 °C; initial oven
temperature, 35 °C; initial time, 2.5 min; program rate, 10 °C/
min; final oven temperature, 250 °C; final time, 10 min.
N-Acetyl N′-ter t-Bu tyl Hyd r a zid e (1). A 1-L three-neck
round-bottom flask equipped with a mechanical stirrer, inert
gas inlet adapter, and rubber septum was charged with 6.24
g of tert-butylhydrazine hydrochloride (50 mmol, 1.0 equiv),
500 mL of dry CH2Cl2, and 15.6 mL of distilled Et3N (11.3 g,
112 mmol, 2.24 equiv). After the suspension was cooled to
approximately -90 °C, a solution of 3.6 mL (3.97 g, 51 mmol,
1.0 equiv) of acetyl chloride in 25 mL of CH2Cl2 was added
with vigorous stirring. The suspension was stirred for 20 min
and then allowed to warm to room temperature. A 90-mL
portion of 1.0 M aqueous KOH (90 mmol, 1.8 equiv) was added
to liberate Et3N from its salt, the CH2Cl2 layer was separated,
and NaCl was added to the aqueous layer, which was then
extracted five times with Et2O. The combined organic layers
were washed with brine and passed through a pad of Na2SO4
and then silica gel. After removal of solvent and Et3N in vacuo,
the product remained as a water-soluble yellow oil that
crystallized with difficulty (4.07 g, 63% yield). Although this
material was sufficiently pure, as ascertained by NMR, for use
in the next step, the hydrazide can be purified by column
chromatography (Rf ) 0.13 in 100% EtOAc). Mp 61-64.5 °C.
(40) Nau, W.; Harrer, H. M.; Adam, W. J . Am. Chem. Soc. 1994,
116, 10972.
(41) Gibian, M. J .; Corley, R. C. Chem. Rev. 1973, 73, 441.
(42) Bartlett, P. D.; McBride, J . M. Pure Appl. Chem. 1967, 15, 89.
(43) Koenig, T.; Fischer, H. In Free Radicals; Kochi, J . K., Ed.; Wiley
& Sons: New York, 1973; Vol. 1, p 157.

Use of Azoalkanes to Generate Captodative Radicals
J . Org. Chem., Vol. 63, No. 26, 1998 9767
NMR shows that the hydrazide exists as two geometrical
isomers because of restricted rotation about the amide link-
age.44 1H NMR (CDCl3): 1.09 (9H, s), 2.00, 2.11 (3H, s) 3.4 (b
s), 4.7 (b s), 6.8 (b s), 6.9 (b s). 13C NMR (CDCl3) (smaller
Z-isomer peaks are in parentheses): 21.09 (19.54), 27.14
(26.94), 54.98 (54.21), 169.33 (177.18).
(CDCl3): 0.27 and 0.29 (9H, s), 1.77 and 1.72 (3H, s). 13C NMR
(CDCl3): 1.16, 1.20, 27.3, 27.7, 92.7, 92.8, 117.0. UV: λmax 368
nm.
Isobu ta n e. 13C NMR (C6D6): δ 23.6, 24.8.
Isobu ten e. 13C NMR (C6D6): δ 24.0, 111.0, 141.8.
2-[(Tr im eth ylsilyl)oxy]p r op en en itr ile (6) was prepared
from pyruvonitrile and N,O-bis(trimethylsilyl)acetamide.45 1
H
N-Acetyl N′-ter t-Bu tyl Dia zen e (2). Under nitrogen, a
2.10 g (16.1 mmol, 1.0 equiv) portion of hydrazide 1 was
dissolved in 80 mL of dry, distilled CH2Cl2 with stirring. The
mixture was cooled to approximately -78 °C, and a solution
of 7.85 g (17.7 mmol, 1.1 equiv) of Pb(OAc)4 in 30 mL of CH2-
Cl2 was added over 5 min with stirring. The orange mixture
was allowed to warm to room temperature and stirred for 1
h. The CH2Cl2 was decanted off, and the residue in the flask
was washed with more CH2Cl2. The combined organic layer
was then washed twice with chilled, saturated NaCl, once with
75 mL of chilled, saturated NaHCO3, and again with 75 mL
of chilled, saturated aqueous NaCl. The solution was dried over
MgSO4, the solid residue was filtered off, and the solvent was
removed in vacuo. Flash distillation into a liquid nitrogen cold
trap removed any nonvolatile impurities, and the moderately
pure yellow oil (1.84 g, 89% yield) was carried on directly to
NMR (C6D6): 0.00 (9H, s), 4.47 (1H, d, J ) 2.3 Hz), 4.52 (1H,
1
d, J ) 2.3 Hz). H NMR (CDCl3): 0.30 (9H, s), 5.03 (d, 1H, J
) 2.3 Hz), 5.10 (1H, d, J ) 2.3 Hz). The reported J of 0.23 Hz
is erroneous, but we used the published 13C shifts to identify
6 in the product mixture from 3. 13C NMR (C6D6): -0.5, 108.9,
116.1, 131.2. lit.45 13C NMR (CDCl3): -0.50, 109.06, 115.91,
130.81.
2-[(Tr im eth ylsilyl)oxy]p r op a n en itr ile (7)46 was pre-
pared from distilled acetaldehyde, trimethylsilyl cyanide, and
catalytic Et3N in dry CH2Cl2 at 0 °C.23 1H NMR (C6D6): 0.00
(9H, s), 1.02 (3H, d, J ) 6.6 Hz), 3.84, (2H, q, J ) 6.6 Hz). 13C
NMR (C6D6): -0.6, 22.8, 57.4, 120.4.
3,3-Dim eth yl-2-[(tr im eth ylsilyl)oxy]-2-cyan obu tan e (8).47
A 0.25-mL portion of pinacolone (200 mg, 1.0 equiv) and 20
mL of CH2Cl2 freshly distilled from CaH2 were placed into a
50 mL flask. Under Ar, the solution was cooled in an ice bath,
and 0.3 mL (220 mg, 1.1 equiv) of TMSCN in 10 mL of CH2Cl2
was added via cannula. A catalytic amount of ZnI2 was added,
and the solution was allowed to warm and stir overnight under
Ar. The 1H NMR spectrum of the crude reaction mixture
showed complete conversion to 8, whose chemical shifts were
correlated with those of the product mixture from decomposi-
1
be used in the next step. H NMR (CDCl3): 1.28 (9H, s), 2.25
(3H, s). 13C NMR (CDCl3): 20.9, 26.1, 69.7, 189.7. TLC: Rf )
0.29 (10% EtOAc/hexane).
N-t-Bu tyl N′-(1-[(Tr im eth ylsilyl)oxy]-1-cya n o)eth yl Di-
a zen e (3). A 1.84-g (14.4 mmol, 1.0 equiv) portion of crude 2
was dissolved with stirring in 70 mL of dry, distilled CH2Cl2
under N2. The solution was cooled to approximately -78 °C,
and 1.71 g of trimethylsilyl cyanide (TMSCN) (17.2 mmol, 1.2
equiv) in 40 mL of CH2Cl2 was added over one min. The Et3N
catalyst (150 mg, 1.4 mmol, 0.1 equiv) was added, and the
mixture was stirred and allowed to warm overnight. After the
reaction was complete as ascertained by TLC (Rf ) 0.39 50%
CH2Cl2/hexane), all volatiles were removed in vacuo. The crude
3 was quite pure but was subjected to column chromatography
on dry silica gel, eluting with dry 50% CH2Cl2/hexane. The
1
tion of 3. H NMR (C6D6): 0.19 (9H, s), 0.90 (9H, s), 1.19 (3H,
s).13C NMR (C6D6): 1.1, 23.7, 24.5, 38.8, 76.3, 121.8.
2,3-Dicya n o-2,3-b is[(t r im et h ylsilyl)oxy]b u t a n e 10a 32
1
(lower Rf). H NMR (CDCl3): 0.3 (18 H, s), 1.59 (3 H, s). 13C
NMR (CDCl3): 0.9, 22.6, 75.1, 119.4. 10b (higher Rf). 1H NMR
(CDCl3): 0.29 (18 H, s), 1.78 (3H, s). 13C NMR (CDCl3): 1.02,
1
24.2, 75.5, 118.8. 10a ,b. H NMR (C6D6): 0.13 (18 H, s), 0.26
(18 H, s), 1.21 (6 H, s), 1.56 (6H, s). 13C NMR (C6D6): 0.90,
22.72, 24.16, 75.38, 75.72, 118.99, 119.66.
1
diazene was a faintly yellow oil. H NMR (CDCl3): 0.24 (9H,
s), 1.27 (9H, s), 1.60 (3H, s), (C6D6) 0.221 (9H, s), 1.06 (9H, s),
1.39 (3H, s). 13C NMR (C6D6): 1.6, 26.2, 27.6, 68.5, 93.5, 118.7.
UV: λmax ) 366 nm, ꢀ ) 22.4.
2-P h en ylth io-2-[(tr im eth ylsilyl)oxy]pr opan en itr ile (12)
was synthesized from pyruvonitrile and phenyltrimethylsilyl
sulfide.34 1H NMR (CDCl3): 0.14 (9H, s), 1.91 (3H, s), 7.37-
7.68 (5H, m).
1-(1,1-Dim e t h yle t h ox y)-2,2,6,6-t e t r a m e t h y lp ip e r i-
d in e (16)48 was prepared by 366-nm irradiation of a freeze-
thaw degassed C6D6 solution of azo-tert-butane in the presence
of 1 molar equiv of TEMPO. The tube was opened, and the
mixture was chromatographed on silica gel to yield 3 mg of
pure material. 1H NMR (CDCl3): 1.29 (6H, s), 1.31 (6H, s),
1.42 (9H, s), 1.5 (6H, m). 13C NMR (C6D6): 17.5, 20.6, 29.6,
35.1, 41.1, 59.3, 77.5.
1-(1-(Tr im eth ylsilyl)oxy-1-cya n oeth oxy)-2,2,6,6-tetr a -
m eth ylp ip er id in e (15) was prepared by allowing an excess
of 5 to decompose overnight in the presence of TEMPO. 1H
NMR (C6D6): 0.25 (9H, s), 1.05 (3H, s), 1.09 (3H, s), 1.22 (3H,
s), 1.36 (3H, s), 1.65 (3H, s), 1.0-1.5 (6H, m). 13C NMR
(C6D6): 1.0, 16.9, 20.5, 20.7, 28.34, 33.4, 33.9, 40.1, 40.2, 59.9,
60.28, 98.16, 118.8. Calcd for C15H30N2O2Si + H+: 299.21548.
Found: 299.21603.
Th er m olysis of 3 w ith P h SH. A dry NMR tube equipped
with a 7/25 joint was charged with 59 mg of 3 (0.26 mmol)
and 51 mg of thiophenol (0.46 mmol, 0.9 equiv) in 1 mL of
C6D6. The tube was freeze-thaw degassed, sealed, and then
heated at 90 °C for 15 h. After the tube was opened its contents
were analyzed promptly by GC/MS. Besides the products
mentioned in the text, unknowns were found at 16.63, 22.32,
and 22.67 min.
Dia cetyl Dia zen e (4).21 To a solution of 1,2-diacetyl
hydrazine (5.0 g, 1.0 equiv) in CH2Cl2 (215 mL) chilled in an
ice-water bath was added Pb(OAc)4 (21.0 g, 1.1 equiv) in CH2-
Cl2 (80.0 mL). The mixture was immediately allowed to warm
to 25 °C, was stirred for 2 h, was recooled to 0 °C, and was
filtered into a chilled separatory funnel. The solid was rinsed
with CH2Cl2, and the solution was quickly washed with chilled
50% brine (150 mL), chilled 50% brine/saturated NaHCO3 (200
mL), and chilled brine once more (150 mL). After drying over
MgSO4, filtration, and removal of solvent in vacuo, the crude
1
product was reasonably pure. Yield: 1.77 g, 36%. H and 13C
NMR spectra showed the only impurity to be acetic anhydride.
Diacetyl diazene is extremely electrophilic and is rapidly
decomposed by water with gas evolution. Attempted nonaque-
ous workups gave higher yields but more impurities. Because
purification by distillation proved futile and chromatographic
media consumed the desired product, 4 was carried on to the
next step without purification. 1H NMR (CDCl3): 2.36 (s), 2.22
(acetic anhydride). 13C NMR (CDCl3): 187.0, 20.4.
N,N′-bis(1-[Tr im eth ylsilyl)oxy]-1-cya n o)eth yl Dia zen e
(5). To a solution of crude diacetyl diazene (∼0.68 g, 1.0 equiv)
contaminated with acetic anhydride (∼0.075 g) in CH2Cl2 (30
mL) cooled to -78 °C was slowly added TMSCN (1.3 mL, 2.0
equiv) in CH2Cl2 (10 mL) followed by the dropwise addition of
Et3N (0.2 mL, 0.25 equiv). After 0.5 h, the bath was removed,
and the reaction was allowed to warm to 0 °C. All volatiles
were removed in vacuo at 0 °C. Cold (ice-water) chromatog-
raphy on very dry silica gel with CH2Cl2/hexanes (20-50%)
provided 1.17 g (63%) of 5 as a roughly 50/50 mixture of
diastereomers that can be stored at <-80 °C. 1H NMR
(45) Barton, D. H. R.; Chern, C.-Y.; J aszberenyi, J . C. Aust. J . Chem.
1995, 48, 407.
(46) Constantieux, T.; Picard, J .-P. Organometallics 1996, 15, 1604.
(47) Evans, D. A.; Carroll, G. L.; Truesdale, L. K. J . Org. Chem.
1974, 39, 914.
(44) Beshan, I. P.; Khrustalev, V. A.; Zelenin, K. N.; Nikolaev, B. P.
J . Org. Chem. USSR 1978, 14, 696.
(48) Kovtun, G. A.; Aleksandrov, A. L.; Golubev, V. A. Izv. Akad.
Nauk SSSR Ser. Khim. 1974, 23, 2197.

9768 J . Org. Chem., Vol. 63, No. 26, 1998
Billera et al.
Th er m olysis of 5 w ith Th iop h en ol. A sealed, degassed
solution of 5 (0.11 mmol) and PhSH (0.22 mmol) in 0.5 mL of
C6D6 was allowed to stand overnight at 40 °C. The contents of
the tube were analyzed by GC comparison with authentic
compounds and by CI/GC/MS. The products, their retention
times, and their relative areas as follows: were 7, 7.78, (176);
PhSH, 10.68, (573); 10a ,b, 17.33, 17.72 (174); 12, 20.16, (57);
PhSSPh, 23.91, (100). Two unknowns, at 25.44 and 25.53 min,
had peak areas of 52 and 56, respectively. The GC conditions
were 0.25 mm × 30 m DB5, 40 °C for 5 min, 10 °C/min to 250
°C with 5 min final time; injector, 180 °C; detector, 250 °C.
NMR TEMP O Toler a n ce Exp er im en t. Solutions of 8, 4,
and 2 mg/mL of TEMPO in CDCl3 were prepared. To 0.5 mL
of each solution was added 0.11 mL (100 mg) of ethyl acetate.
All three samples exhibited 1H NMR spectra with some line
broadening, but the 8 mg/mL spectrum was so distorted that
the ethyl-group coupling was difficult to observe. A quantita-
tive 13C NMR experiment (see below) on the most concentrated
solution gave sharp peaks with the proper area ratios, indicat-
ing that 13C integrals should be reliable at TEMPO concentra-
tions below 8 mg/mL.
P h otolysis of 3 w ith TEMP O. To a dry 5-mm NMR tube
equipped with a 7/25 joint was added 127 mg of azoalkane 3
(0.56 mmol) and 90 mg of TEMPO (0.58 mmol) as a solution
in 0.6 mL of C6D6. The orange-red solution was freeze-thaw
degassed three times in liquid nitrogen and then sealed. The
sample was irradiated at 366 nm and 21 °C with periodic
manual agitation until the orange-red color of the TEMPO had
been reduced to an intensity corresponding to 4-8 mg/mL, as
judged by visual comparison to standard solutions. The proton
and nonquantitative carbon NMR spectra were acquired to
determine the spectrometer and decoupler offset. A delay of
20 s was established between pulses to allow for complete
relaxation. The number of data points was set at the maximum
of 128K for accurate line-shape definition, and the spectral
width was set at 160 ppm to encompass the entire spectrum.
The experiment was performed with gated decoupling to
eliminate NOE. After a total of 1922 scans were acquired, the
FID was processed with slight line broadening (0.5 Hz) to
reduce noise, and, was then phased. The chemical shifts were
referenced to C6D6 at 128.0 ppm. By comparison with the 13C
spectra of the authentic products, nearly all of the peaks were
assigned. The peaks were integrated in windows of about 10-
20 ppm, so that accurate baseline corrections could be applied.
Correcting the integrals for multiple carbons led to numbers
proportional to the concentration of products. Table 3 sum-
marizes the chemical shifts, areas, total area for each com-
pound, and percent yield.
hydrazine were added. The orange-red color of TEMPO disap-
peared, but the tube retained a slight yellow tint due to excess
3. The NMR tube was capped with a rubber septum, and a
conventional 13C spectrum was acquired. No new peak in the
window near 18 ppm appeared, indicating that the previously
seen peaks corresponded to 17.
Decom p osition of 5 w ith TEMP O. To an NMR tube
equipped with a 7/25 joint and a sealing constriction was
added 20 mg of TEMPO (0.13 mmol) and 11 mg of 5 (0.035
mmol) in 0.4 mL of C6D6. Because 5 is extremely water and
heat sensitive, anhydrous techniques were employed while
keeping the sample under 10 °C. The tube was freeze-thaw
degassed as described above and sealed. After the solution had
stood at 25 °C for 1 day, the color had faded noticeably, but a
conventional 13C spectrum revealed that no 17 was present.
Th er m olysis k in etics of 3 a n d 5 were carried out on a
constant-volume, variable-pressure gas-evolution apparatus.24
The rate constants for ∼0.25 M 3 in Ph2O were (temp (°C),
104k (s-1), method) 74.9, 0.582 ( 0.003, I; 80.0, 1.11 ( 0.003,
I; 90.0, 3.85 ( 0.016, I; 95.1, 7.38 ( 0.076, G; while those for
0.2 M 5 in PhCH3 were 24.9, 0.664 ( 0.001, I; 29.3, 1.41 (
0.02, G; 29.2, 1.37 ( 0.004, I; 33.7, 2.46 ( 0.01, I; 39.4, 4.83 (
0.01, I. “I” indicates that the experimental infinity point was
used while “G” signifies the Guggenheim method.
Ca ge Effects. The cage effect of 5 was determined using a
hexane solution of 0.0846 M TEMPO and <0.04 M 5. The
solution was kept cool to prevent azoalkane thermolysis as it
was degassed and sealed into a Pyrex tube attached to a
cuvette. The absorbance of TEMPO was determined with the
cool solution; then this solution was allowed to decompose at
approximately 40 °C for several hours. The final absorbance
of TEMPO was determined, and the amount of nitrogen
evolved was measured on a vacuum line and To¨pler pump.
The nitrogen yield gave the amount of 5 that decomposed
(0.126 mmol), while the change in TEMPO absorbance counted
the escaped radicals (0.215 mmol). On the basis of these
numbers, the efficiency was 85%, and the cage effect was 15%.
The same technique was employed for azo-t-octane, but, since
this compound is thermally stable,51 it was not necessary to
keep the solution cool.
Ack n ow led gm en t. We thank the National Science
Foundation and the Robert A. Welch Foundation for
support of this work.
Su p p or tin g In for m a tion Ava ila ble: Table 3, a tabula-
tion of 13C NMR chemical shifts and areas for all products of
the photolysis of 3 in the presence of Tempo (2 pages). This
material is contained in libraries on microfiche, immediately
follows this article in the microfilm version of the journal, and
can be ordered from the ACS; see any current masthead page
for ordering information.
N-Hyd r oxy-2,2,6,6-tetr a m eth ylp ip er id in e (17) was gen-
erated by reduction of TEMPO with phenylhydrazine.49 Car-
bons 2, 3, and 4 appear at 18.04, 40.42, and 59.18 ppm,
respectively, but the methyl groups could not be seen on
account of rapid conformational interconversion.50 To confirm
the presence of 17 from thermolyzed 3, the NMR tube was
opened and 5 mg of TEMPO and a slight excess of phenyl-
J O981372H
(49) Lee, T. D.; Keana, J . F. W. J . Org. Chem. 1975, 40, 3145.
(50) Anderson, J . E.; Corrie, J . E. T. J . Chem. Soc., Perkin Trans. 2
1992, 1027.
(51) Bandlish, B. K.; Garner, A. W.; Hodges, M. L.; Timberlake, J .
W. J . Am. Chem. Soc. 1975, 97, 5856.