Estimation of Bond Dissociation Energies and Radical Stabilization Energies by ESR Spectroscopy

Article abstract of DOI:10.1021/jo971940d

Correlations of various indices of the stability and reactivity of carbon-centered radicals with ESR hyperfine splitting constants have been examined. For a large number of mono- and disubstituted radicals there is a moderately good linear correlation of α-proton hyperfine splitting constants (a(H_α)) with radical stabilization enthalpies (USE) and with BDE(C-H), the C-H bond dissociation energies for the corresponding parent compounds determined from thermodynamic and kinetic studies of C-C homolysis reactions. There is a similarly satisfactory linear correlation of a(H_α) with BDE-(C-H) determined by Bordwell's electrochemical and acidity function method. In all cases the correlations fail for nonplanar radicals. As expected, β-proton hyperfine splitting constants (a(H_βMe)) for radicals with a freely rotating methyl substituent are less sensitive to deviations from planarity and give better linear correlations with RSE and BDE(C-H). The correlations cover a range of more than 20 kcal/mol and are reliable predictors of RSE and BDE(C-H) for a variety of radicals including captodative species. However, the correlations fail for significantly nonplanar radicals and for radicals with cyclic delocalized systems, e.g., cyclopentadienyl. The ratio a(H_βMe)/ a(H_α) for suitably substituted radicals provides an index of pyramidalization and allows one to decide for which compounds values of RSE and BDE(C-H) can be confidently estimated.

Full text of DOI:10.1021/jo971940d

J . Org. Chem. 1998, 63, 1935-1943
1935
Estim a tion of Bon d Dissocia tion En er gies a n d Ra d ica l
Sta biliza tion En er gies by ESR Sp ectr oscop y
†
†
,‡
J ochen J . Brocks, Hans-Dieter Beckhaus, Athelstan L. J . Beckwith,* and
Christoph R u¨ chardt*,
†
Institut f u¨ r Organische Chemie and Biochemie der Universit a¨ t Freiburg, Albertstrasse 21,
D-79104 Freiburg i. Br., Germany, and Department of Chemistry, Australian National University,
Canberra 0200, Australia
Received October 21, 1997
Correlations of various indices of the stability and reactivity of carbon-centered radicals with ESR
hyperfine splitting constants have been examined. For a large number of mono- and disubstituted
R
radicals there is a moderately good linear correlation of R-proton hyperfine splitting constants (a(H ))
with radical stabilization enthalpies (RSE) and with BDE(C-H), the C-H bond dissociation energies
for the corresponding parent compounds determined from thermodynamic and kinetic studies of
C-C homolysis reactions. There is a similarly satisfactory linear correlation of a(H
C-H) determined by Bordwell’s electrochemical and acidity function method. In all cases the
correlations fail for nonplanar radicals. As expected, â-proton hyperfine splitting constants
a(H Me)) for radicals with a freely rotating methyl substituent are less sensitive to deviations
R
) with BDE-
(
(
â
from planarity and give better linear correlations with RSE and BDE(C-H). The correlations cover
a range of more than 20 kcal/mol and are reliable predictors of RSE and BDE(C-H) for a variety
of radicals including captodative species. However, the correlations fail for significantly nonplanar
radicals and for radicals with cyclic delocalized systems, e.g., cyclopentadienyl. The ratio a(H
â
Me)/
a(H ) for suitably substituted radicals provides an index of pyramidalization and allows one to
R
decide for which compounds values of RSE and BDE(C-H) can be confidently estimated.
2
In tr od u ction
Early work on the factors affecting the magnitudes of
R â
a(H ) and a(H ) showed that for planar carbon-centered
radicals
Radical stabilization energies and bond dissociation
enthalpies (BDE) are useful for the prediction or inter-
pretation of reactivity and selectivity in free radical
chemistry. Values of stabilization energy or of BDE(C-
H) cannot be measured directly but must be deduced
indirectly from the results of thermochemical, kinetic, or
electrochemical experiments. Thus the quality and
comparability of values measured by different techniques
depend both on the accuracy of the experimental data
and on their interpretation. Consequently, it is some-
times difficult to assess the validity of a new method or
even the accuracy of a single value in a series obtained
by similar measurements. This is particularly so if no
appropriate comparable data are available in the litera-
ture.
a(H ) ) Q F(C )
(1)
(2)
R
R
R
a(H Me) ) Q F(C )
â
â
R
where F(C
R
) is the spin density on the radical center, Q
are constants, and a(H Me) is the â-proton
R
and Q
â
â
hyperfine splitting constant for a freely rotating methyl
group attached directly to the radical center. On the
basis of eqs 1 and 2, it should be possible to use either
R â
a(H ) or a(H Me) to determine values of R-spin density,
and hence the degree of delocalization of the free electron.
However, the development of quantitative relationships
between hyperfine coupling constants and radical stabi-
lization or BDE is complicated by the fact that eqs 1 and
1
In a recent paper, we examined the relationship of
radical stabilization enthalpy (RSE) to the R-proton ESR
R
hyperfine splitting constant (a(H )) for 15 different
2
are not valid for nonplanar radicals. Deviations from
planarity due to strain, such as occurs for cyclopropyl and
radicals including unsaturated and captodative species.
The values of RSE ranged over 22 kcal/mol and of a(H
3
other small cyclic radicals, or from the effect of elec-
R
)
3
4
tronegative R-substituents, e.g., halogen, oxygen, or
over 14 G. A good linear correlation was observed. In
the present study, we extend this approach to further
examples and show that it is possible to use both R- and
â-proton hyperfine splitting constants to predict values
of RSE and BDE(C-H) and to estimate the consistency
of a series of data.
5
R â
nitrogen, lower the values of Q and Q . Nevertheless,
there have been a number of earlier attempts to correlate
stabilization energies or bond dissociation energies with
R â R
a(H ) or a(H Me). Most such studies have involved a(H ).
6
In one of the earlier papers, Green and Walton
ESR
reported that stabilization energies (SE ) derived ex-
†
Universit a¨ t Freiburg.
Australian National University.
‡
(4) Dobbs, A. J .; Gilbert, B. C.; Norman, R. O. C. J . Chem. Soc. A
l971, 124-135.
(5) Armstrong, D. A.; Rank, A.; Yu, D. J . Am. Chem. Soc. 1993, 115,
666-673; J . Am. Chem. Soc. 1994, 116, 8432. Shaffer, S. A.; Turecek,
F.; Ceruy, R. L. J . Am. Chem. Soc. 1993, 115, 12117-12124.
(6) Green, I. G.; Walton, J . C. J . Chem. Soc., Perkin Trans. 2 1984,
1253-1257.
(
1) Welle, F. M.; Beckhaus, H.-D.; R u¨ chardt, C. J . Org. Chem. 1997,
2, 552-558.
2) Fischer, H. Z. Naturforsch. 1964, 19a, 855-867. Fischer, H. Z.
Naturforsch. 1965, 20a, 428-432.
3) For a discussion of factors affecting deviations from planarity of
radicals including the cyclopropyl and oxiranyl radicals, see ref 4.
6
(
(
S0022-3263(97)01940-3 CCC: $15.00 © 1998 American Chemical Society
Published on Web 02/27/1998

1
936 J . Org. Chem., Vol. 63, No. 6, 1998
Brocks et al.
clusively from experimental ESR rotational barriers gave
a linear relationship when plotted against log(a(H )),
where a(H ) is the proton hyperfine splitting for the
reactivity. In the present work we have compared cor-
R
relations of a(H ) and of a(H Me) with BDE(C-H) and
R â
16
R
with R u¨ chardt’s radical stabilization enthalpies (RSE).
terminal methylene groups of polyenyl radicals. In
7
contrast, Nicholas and Arnold found a linear relationship
Resu lts a n d Discu ssion
ESR
between a(H
R
) and SE
for similar species. Nonhebel
and Walton8 examined the relationship between the
rotational barriers and a(H ) for some saturated radicals
In d ices of Ra d ica l Sta bility a n d Rea ctivity. Al-
though it is clear that a decrease in R-spin density
represents delocalization of the free electron, it is not
axiomatic that different mechanisms of spin distribution,
such as π-delocalization, hyperconjugation, spin polariza-
tion, and anomeric interactions, will each lead to the
same increase in radical stabilization for the same degree
of spin delocalization.17 Furthermore, there are other
factors that may influence the stabilization energy as
estimated from ∆BDE(C-H) or ∆BDE(C-C). They
include interactions between substituents such as van
der Waals repulsions, dipolar effects, hydrogen bonding,
and hydrophobic interactions, all of which contribute to
the difference in strain enthalpy between a radical and
its precursor. To minimize the importance of such effects,
we have used radical stabilization enthalpy (RSE) defined
as the difference in stability between a purely hydrocar-
R
9
and found a linear correlation, while Leroy et al. showed
that there also exists a linear relationship between the
F(C ) of six captodative radicals and their relative
R
stabilization energies determined by ab initio calcula-
tions. Other examples of attempts to correlate hyperfine
splitting constants with factors related to stabilization
energies include Stein’s linear correlations of the kineti-
cally determined BDE(C-C) of substituted ethylbenzenes
with the benzylic hyperfine splitting constant10 and of
the BDE(C-O) of substituted anisoles with the hyperfine
splitting constants for the ortho and para protons in
substituted phenoxy radicals.11
For aromatic compounds, Arnold et al.12 have published
extensive lists of substituent constants σ
and para-substituted benzyl radicals. σ
•
R
•
(X) for meta-
R
(X) is defined
•
•
by σ
hyperfine splitting of the benzylic proton in the substi-
tuted benzyl radical and a(H (0)) is the corresponding
splitting for the benzyl radical. The σ
R
(X) ) 1 - a(H
R
(X))/a(H
R
(0)), where a(H
R
(X)) is the
bon radical CXYZ (X, Y, Z ) H or alkyl) and the
•
corresponding substituted radicals CXYZ (X ) H or alkyl;
Y ) H, alkyl or functional group; Z ) functional group),
i.e., the difference in stability between a purely hydro-
carbon radical and its analogue in which one or more of
the alkyl groups have been replaced by substituents. One
of the advantages of the use of RSE’s is that they avoid
the problem that values of differences in BDE(C-H)
clearly do not correspond accurately to the differences
in stability between primary, secondary, and tertiary
alkyl radicals. Extensive compilations of values of RSE
based on thermochemical and kinetic data for the ho-
molysis of C-C bonds are available.16
R
•
R
scale gives good
linear correlations with the rates of various radical
reactions in Taft equations. Similar results involving
extensive tables of σ have been reported by J ackson et
•
1
3
•
al.
Very recently a σ scale based on the ESR D
parameter for substituted 1,3-arylcyclopentane-1,3-diyl
radicals has been published.14
As Q
the radical center than is Q
values of a(H Me) would give a better correlation with
indices of radical stabilization than do values of a(H
â
(eq 2) is less diminished by pyramidalization of
R
(eq 1), one might expect that
â
2
,4
).
Although Norman et al. had earlier noted that there is
a good correlation of a(H Me) with F(C ), the only
correlation of a(H Me) with radical reactivity appears to
Despite the fact that they include contributions arising
from the release of strain in the precursor molecule and
differences between other nonbonded interactions in the
precursor and the radical, values of BDE(C-H) for the
parent compound are widely used to estimate the relative
stabilities and reactivities of radicals. McMillen and
R
4
â
R
â
be that of Afanas’ev who developed a Taft equation for
the rates of radical substitution reactions.15 He found
•
18
that there is a good linear correlation between σ and
Golden’s early review includes extensive lists of BDE-
•
a(H
â
Me) for radicals of the type X(CH
3
)HC .
(C-H) values taken from the literature, while more
recently Tsang19 has published a selection of “best”
values. An excellent self-consistent series of BDE(C-
H) has been experimentally determined by Bordwell
using a combination of electrochemical and acidity func-
tion measurements.20 Values of BDE(C-H) are also
Despite the undoubted utility of some of the correla-
tions outlined above, there remains the question of which
type of hyperfine splitting gives the best correlations with
consistent and reliable indicators of radical stability and
(
7) Nicholas, A. M. de P.; Arnold, D. R. Can. J . Chem. 1986, 64,
2
70-276.
(17) For a comment, see: Sustmann, R.; Korth, H. G. Adv. Phys.
Org. Chem. 1990, 26, 131-178.
(8) Nonhebel, D. C.; Walton, J . C. J . Chem. Soc., Chem. Commun.
1
984, 731-732.
(18) McMillen, D. F.; Golden, D. M. Annu. Rev. Phys. Chem. 1982,
32, 493-532.
(9) Leroy, G.; Peeters, D.; Sana, M.; Wilante, C. Bull. Soc. Chim.
Belg. 1988, 97, 1003-1010.
(19) Tsang, W. In Energetics of Organic Free Radicals; Somoes, A.
M., Greenberg, A., Liebman, F., Eds.; Black Academic and Profes-
sional: London, 1996; pp 22-58.
(
10) Suryan, M. M.; Stein, S. E. J . Phys. Chem. 1989, 93, 7362-
7
365.
(11) Suryan, M. M.; Kafafi, S. A.; Stein, S. E. J . Am. Chem. Soc.
(20) (a) Bordwell, F. G.; Zhang, X.-M. Acc. Chem. Res. 1993, 26, 510-
517. (b) Bordwell, F. G.; Zhang, X.-M.; Alnajjar, M. S. J . Am. Chem.
Soc. l992, 114, 7623-7629. (c) Bordwell, F. G.; Harrelson, J . A.; Zhang
X. J . Org. Chem. 1991, 56, 4448-4450. (d) Bordwell, F. G.; Zhang,
X.-M.; Filler, R. J . Org. Chem. 1993, 58, 6067-6071. (e) Bordwell, F.
G.; Satish, A. V. J . Am. Chem. Soc. 1994, 116, 8885-8889. (f) Bordwell,
F. G.; Gallagher, T.; Zhang, X. J . Am. Chem. Soc. 1991, 113, 3495-
3497. (g) Bordwell, F. G.; Cheng, J . P.; Harrelson, J . A. J . Am. Chem.
Soc. l988, 110, 1229-1231. (i) Bordwell, F. G.; Harrelson, J . A.; Satish,
A. V. J . Org. Chem. l989, 54, 3101-3105. (j) Bordwell, F. G.; Cheng,
J .-P.; J i, G.-Z.; Satish, A. V.; Zhang, X. J . Am. Chem. Soc. 1991, 113,
9790-9795.
1
996, 118, 775-778.
(
12) Dust, J . M.; Arnold, D. R. J . Am. Chem. Soc. 1983, 105, 1221-
1
1
6
227. Wayner, D. D. M.; Arnold, D. R. Can. J . Chem. 1984, 62, 1164-
168. Wayner, D. D. M.; Arnold, D. R. J . Am. Chem. Soc. l983, 105,
531.
(
13) J ackson, R. A.; Sharifi, M. J . Chem. Soc., Perkin Trans. 2 1984,
253-1257 and references therein.
14) Adam, W.; Harrer, H. M.; Kita, F.; Korth, H.-G.; Nau W. M. J .
1
(
Org. Chem. 1997, 62, 1419-1426.
(
(
15) Afanas’ev, B. Int. J . Chem. Kinetics l975, 7, 857-877.
16) R u¨ chardt, C.; Beckhaus, H.-D. Top. Curr. Chem. 1985, 130,
1
-22. For more recent references to this method for determining RSE
and BDE(C-H), see ref 1.
(21) (a) Pasto, D. J .; Krasnansky, R.; Zercher, C. J . Org. Chem. 1987,
52, 3062-3072. (b) Pasto, D. J . J . Org. Chem. 1988, 53, 8164-8175.

BDE Estimation by ESR Spectroscopy
J . Org. Chem., Vol. 63, No. 6, 1998 1937
•
Ta ble 1. RSE, BDE(C-H), ESR P a r a m eter s, a n d P yr a m id a liza tion Qu otien ts for th e Ra d ica ls CXYZ
RSEa/
kcal mol
BDE(C-H) /
b
-
1
-1
Pt c
Psec d
no.
X
Y
Z
kcal mol
aR/G
a(HâMe)/G
1
1
2
2
3
3
4
4
5
5
6
6
7
7
8
8
9
9
0a
0b
1a
1b
2a
2b
3
4a
4b
5
a
b
a
b
a
b
a
b
a
b
a
b
a
b
a
b
a
b
Me
Me
Ph
Ph
OPr
OBu
NH2
NH2
COPh
COPh
CO2Et
CO2Et
CN
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Ph
Me
H
Me
H
Me
H
Me
H
Me
H
Me
H
Me
H
Me
H
Me
H
Me
H
Me
H
Me
H
H
Me
H
H
Me
H
≡ 0
95.7e
98.7e
87.3
90.3
93.9
96.9
94.2
97.1
89.9
92.9
92.6
95.6
91.9
94.9
85.6
88.6
84.7
87.7
84.6
87.6
87.3
90.3
89.4
92.4
84.9
81.3
84.3
78.3
82.8
85.8
83.5
79.7
82.7
77.6
80.6
83.1
86.1
22.71f
24.7f
16.5f
≡ 0
22.1f
16.25h
13.5k
14.7k
18.7k
20.48j
20.3f
14.8q
18.0j
1.03
1.02
1.45
1.22
1.04
1.05
1.01
1.12
1.16
1.12
1.09
1.6
-8.4g
g
h
-8.4
17.69
i
i
j
-5.9
19.37
t
i
k
-5.9
21.6
17.9
l
j
-3.9
l
20.7k
-3.9
1.41
1.17
1.21
1.13
m
19.13k
-6.0
-6.0m
21.86
21.56
k
-2.8m
f
m
j
-2.8n
24.68
k
-3.4
20.6
CN
-3.4n
23.1f
o,p
16.60q
CO2Et
CO2Et
COMe
COMe
CN
CN
CN
CN
-10.1
o,p
Ph
-10.1r
-11.4
-11.4r
-2.5n
COMe
COMe
CN
CN
OH
21.52j
20.64j
17.88f
17.8j
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
n
19.22j
18.4k
-2.5
i
-8.1
0.97
1.04
i
OMe
-8.1
s,p
-CONBuCHMeCONBu-
-CONBuCH2CONBu-
CN
CO2Bu
CO2Me
COPh
Ph
-6.3
-6.3
s,p
j
17.1
t
14.9k
NH2
NH2
NH2
NMe2
Ph
-13.8
t
-14.8u
13.2v
15.1f
1.02
1.03
-14.8u
13.0v
w
x
-21.6
-12.9y
-12.9y
8.5
6a
6b
7
8a
8b
9a
9b
0a
0b
1
Ph
Ph
14.7f
13.4g
y
9,10-dihydroanthryl-9-yl
9-methylfluoren-9-yl
fluoren-9-yl
9-methylxanth-9-yl
xanth-9-yl
-15.2
-16.0z
14.56j
12.2f
13.8f
1.05
0.96
0.99
z
13.9f
12.7f
14.0f
11.9f
-16.0
-18.1y
-18.1y
y
CHdCH2
CHdCH2
Me
Me
Me
H
-12.6
-12.6y
-13.7y
-15.7y
15.2
f
1.09
aa
6.4f
pentamethylcyclopentadienyl
inden-1-yl
82.0
83.0
80.9
2
3
bb
bb
anthron-10-yl
-17.8
12.12
a
•
b
Reference 16. In most cases RSE is the mean value of RSE for several radicals CXYZ with different alkyl substituents. Reference
t
Me
•
•
d
sec
) a(Hâ (XHMeC ))/a(HR(XHMeC )). Reference 19. f Reference 23a. g Kratt, G.; Beckhaus,
Me
•
•
e
1
. c P ) a(Hâ (XYMeC ))/a(HR(XYHC )).
P
h
H.-D.; R u¨ chardt, C. Chem. Ber. 1984, 117, 1748-1764. Arnold, D. R.; de P. Nicholas, A. M.; Young, K. M. Can. J . Chem. 1986, 64,
i
j
7
69-772. Birkhofer, H.; Beckhaus, H.-D.; Peters, K.; von Schnering, H. G.; R u¨ chardt, C. Chem. Ber. 1993, 126, 1693-1699. This work.
l m
k
Reference 23b. Sch u¨ le, U. Ph.D. Thesis, University of Freiburg, 1992, ref 1. Gleissner, R. Ph.D. Thesis, University of Freiburg, 1988,
ref 16. Pakusch, J .; Beckhaus, H.-D.; R u¨ chardt, C. Chem. Ber. 1991, 124, 1191-1198. o Rausch, R. Ph.D. Thesis, University of Freiburg,
n
p
1
984. Schulze, R.; Beckhaus, H.-D.; R u¨ chardt, C. J . Pract. Chem. 1990, 332, 325-330. Geminal interactions in the saturated radical
q
r
precursor are neglected. Korth, H.-G.; Lommes, P.; Sustman, R.; Sylander, L.; Stella, L. New. J . Chem. 1987, 11, 365-375. N o¨ lke, M.;
s
Verevkin, S. P.; Beckhaus, H.-D.; R u¨ chardt, C. Liebigs Ann. 1995, 41-51. Brocks, J . J .; Welle, F. M.; Beckhaus, H.-D.; R u¨ chardt, C. in
t
u
preparation. Welle, F. M.; Verevkin, S. P.; Beckhaus, H.-D.; R u¨ chardt, C. Liebigs Ann. 1997, 155-163. Schulze, R.; Beckhaus, H.-D.;
v
R u¨ chardt, C. Chem. Ber. 1993, 126, 1031-1038. MacInnes, I.; Walton, J . C.; Nonhebel, D. C. J . Chem. Soc., Perkin Trans. 2 1987,
x
y
1
789-1794. w Reference 1. Welle, F.; Verevkin, S. P.; Keller, M.; Beckhaus, H.-D.; R u¨ chardt, C. Chem. Ber. 1994, 127, 697-710. Herberg,
z
C. Ph.D. Thesis, University of Freiburg, 1992. Rakus, K.; Sch a¨ tzer, J .; Verevkin, S. P.; Beckhaus, H.-D.; R u¨ chardt, C. Chem. Ber. 1994,
aa
bb
1
27, 1095-1103. Strain enthalpy is implied. Wurche, F. Diploma Thesis, University of Freiburg, 1996.
available from R u¨ chardt’s kinetic and thermochemical
studies.16 Also, a variety of theoretical approaches at
study were obtained from the literature.23 Although the
spectra were recorded in a wide range of solvents, our
analysis of more than 200 hyperfine splitting constants
showed that the values of a(H ) and a(H Me) for any
2
1a
various levels of theory, e.g., UHF or ROHF 4-31G,
Density Functional Theory,22a and MP4 methods, have
22b
R
â
been used to obtain values of BDE(C-H).
single radical vary within a range of <0.3 G if measured
in different nonpolar solvents. Measurements in aqueous
solutions show variations of up to 0.5 G from those made
in nonpolar solvents, while splittings obtained from solid
samples show much larger variability, and hence are not
suitable.
ESR Hyp er fin e Cou p lin g Con sta n ts. Many of the
R â
values of a(H ) and of a(H Me) required for the present
(22) (a) J ursic, B. S.; Timberlake, J . W.; Engel, P. S. Tetrahedron
Lett. 1996, 37, 6473-6474. (b) Leroy, G.; Sana, M.; Wilante, C. J . Mol.
Struct. (Theochem.) 1991, 228, 37-45.
(23) (a) Berndt, A.; Fischer, H.; Paul, H. In Magnetic Properties of
R â
Some values of a(H ) and a(H Me) in nonpolar solvents
Free Radicals; Fischer, H., Hellwege, K.-H., Eds.; Landolt-B o¨ rnstein,
Numerical Data and Functional Relationships in Science and Technol-
ogy, New Series, Volume 9, Subvolume b; Springer-Verlag: Berlin,
required for the present study were not available and
were therefore measured in these laboratories. The
radicals 3a , 4a , 9a , 9b, 10a , 10b, 12a , 12b, and 27 (the
structures of radicals 1-23 are given in Table 1) were
each generated in the cavity of the ESR spectrometer by
1
977. (b) Neugebauer, F. A. Magnetic Properties of Free Radicals;
Fischer, H., Ed.; Landolt-B o¨ rnstein, Numerical Data and Functional
Relationships in Science and Technology, New Series, Volume 17,
Subvolume b; Springer-Verlag: Berlin, 1987.

1
938 J . Org. Chem., Vol. 63, No. 6, 1998
Brocks et al.
precursor because of the adjacent radical center. Hence
2
4 undergoes rapid enolization to form the symmetrical
radical 25.
As the preferential formation of 25 could be reasonably
attributed to polar effects, the experiment was repeated
in the presence of the polarity reversal catalyst tri-
methylamine-borane complex.24 As expected, the inter-
•
mediate nucleophilic Me
3
N‚BH
2
radical abstracted hy-
drogen with high selectivity from the electrophilic 3-po-
sition of the precursor to afford only 9a , the spectrum of
which appeared as a clean quartet (a(H) ) 21.5 G),
although the spectral simulation showed that it was
necessary to include a very small septet coupling (a(H)
)
0.1 G) to account for the line width. The same method
involving catalysis by Me was effective for gen-
3
N‚BH
3
erating the radical 9b from the parent dione.
An attempt to form the radical 28 gave what appears
to be another example of polar effects on hydrogen-atom
abstraction. When the precursor 26 in di-tert-butyl
peroxide was subjected to UV irradiation, the ESR
spectrum with hfc’s a(1H) ) 1.34 G, a(2H) ) 14.30 G,
a(3H) ) 4.04 G, and a(1N) ) 7.00 G was assigned to the
radical 27. Apparently the selectivity of the reaction is
affected not only by statistical factors but also by the
preference of the electrophilic Bu O radical for attack on
the methyl C-H bonds adjacent to the electron-donating
nitrogen rather than on the 2-CH position, even though
the latter would afford the more stabilized product 28.
F igu r e 1. ESR spectrum of radicals 25 and 9a and its
simulation (below); the arrows indicate the quadruplet of 9a .
UV irradiation of a solution of the corresponding C-H
compound and di-tert-butyl peroxide in a nonpolar sol-
vent. Similar irradiation of a mixture of di-tert-butyl
peroxide, triethylsilane (or hexamethylditin), and ethyl
t
•
2
-bromopropionate afforded the l-(ethoxycarbonyl)ethyl
•
radical 6b via abstraction of the bromine atom by Et
or Me
3
Si
•
3
Sn . The 9-methylfluoren-9-yl radical 18a was
generated by thermal homolysis of the dimer in diphenyl
ether. Malononitrile and methylmalononitrile, the cho-
sen precursors for the radicals 10a and 10b, were
insufficiently soluble in nonpolar solvents to give suitable
ESR spectra. However, satisfactory signals were ob-
tained by irradiation of saturated solutions of the two
precursors in a 1:1 mixture of benzene and di-tert-butyl
peroxide, even though in the case of methylmalononitrile
a two-phase system was formed.
3
-Methyl-2,4-pentanedione, when irradiated in di-tert-
butyl peroxide solution, gave a complex ESR signal
comprised of two overlapping spectra (Figure 1). The
major spectrum (80%), a doublet (a(H) ) 1.46 G) of
triplets (a(H) ) 16.25 G) of septets (a(H) ) 3.23 G), was
tentatively assigned to the radical 25, while the minor
component (20%) was a quartet (a(H) ) 21.5 G) assigned
to the radical 9a .
The hyperfine splitting constants of 9a , 25, 27, and the
other radicals generated in the course of this work were
all determined by computer simulation.
Cor r ela tion of Ra d ica l Sta biliza tion En th a lp y
r â
w ith a (H ) a n d a (H Me). Table 1 gives a list of values
of RSE for 40 radicals determined by the R u¨ chardt
method from changes in the values of BDE(C-C) by
comparison with alkyl-substituted ethanes with the same
We suggest that formation of 25 initially involves
t
•
hydrogen-atom abstraction by the electrophilic Bu O
radical from the 3-methyl substituent, the C-H bonds
of which are the least affected by the electron-withdraw-
ing nature of the carbonyl groups. The â-proton in the
species 24 so formed is more acidic than it is in the
(
24) Paul, V.; Roberts, B. P.; Willis, J . J . Chem. Soc., Perkin Trans.
1989, 1953-1261. Paul, V.; Roberts, B. P. J . Chem. Soc., Chem.
Commun. 1987, 1322-1324.
2

BDE Estimation by ESR Spectroscopy
J . Org. Chem., Vol. 63, No. 6, 1998 1939
Ta ble 2. BDE(C-H) for H-CHXY Deter m in ed by th e
Bor d w ell Meth od a n d a (Hr) for th e Cor r esp on d in g
Ra d ica ls
a
BDE(C-H) /
-
1
no.
X
Y
kcal mol
a(HR)/G
98.3b
20.15
19.7
20.95
c
3
3
4
5
6
7
8
9
0
1
9
2
3
4
CO2Me
COMe
COBu
COMe
SR
COPh
COPh(p-OMe)
COPh(p-CN)
-CO2CMe2OCO-
COMe
-COCH2CH2CH2CO-
NO2
CN
CN
COEt
Ph
COPh
CO2Me
H
H
CO2Et
H
H
H
H
b
d
3
3
3
3
3
3
4
4
97
t
e
d
96.3
96f
18.96
c
g
d
96
96
96
96
95.6
95
95
94.4
94
93
91.2
91
91
91
91
90.0
16.9
19.5
c
d
c
c
19.25
c
c
17.5
b
h
19.30
i
k
b
COMe
18.0
i
d
4
4
4
1
4
46
5
4
4
20.1
i
d
Me
21
f
c
CO2Et
CN
Me
H
Me
19.80
j
k
0b
5
19.22
e
d
18.6
c
d
F igu r e 2. Plot of RSE against a(H
R
) showing the line of eq 3
16.30
18.70
i
c
b
in Table 4: (4) outliers for the calculation of eq 3. Compound
numbers are shown in Table 1.
l
d
7
8
-COCH2CH2CH2-
-COCH2CH2CH2CH2-
SEt
17.96
i
d
18.1
15.0
15.1
m
c
strain dissociation enthalpy.16 A consequence of this
49
CO2Et
CN
CN
m
c
5
5
5
53
8
54
5
1
18b
56
1
2
1
1
0
1
2
SMe
89
method is that all saturated acyclic tertiary and second-
ary radicals are defined as having RSE ) 0. Further-
more, similarly substituted radicals, whether secondary
or tertiary, all have the same value of RSE, e.g.,
t
m
c
S Bu
88.9
15.02
88.8n
d
CHdCH2
SPh
Ph
COPh
Ph
Ph
H
14.35
m
c
CO2Et
CO2Et
Ph
CN
Ph
88.8
87.0
85.5
84.9
15.0
b
o
p
q
b
14.94
14.04
15.29
c
•
1
2
•
3
CR R COPh and CHR COPh have RSE ) -6.0 kcal/mol
no matter what the structures of the saturated alkyl
j
5
6b
r
d
84.8
14.7
1
2
3
substituents R , R , and R are. Hence, the values of RSE
given in Table 1 are not necessarily those determined
for the particular radical specified but are the mean
values obtained for a number of radicals of the same type.
However, to maintain consistency, ESR data are given
only for those radicals in which the saturated alkyl
substituent(s) is methyl. Also it should be noted that
fluorene (9-H)
83c
13.9d
r
d
d
cyclopentadiene (1-H)
dihydroanthracene (9-H)
indene (9-H)
81.8
6.06
h
13.4s
7
2
9b
5
81
81r
11.9
d
j
xanthene (9-H)
78.5
75
12.73
i
t
COPh
NMe2
phenalen (1-H)
8.3
67j
6.3d
57
a
The BDE(C-H) data in this table is 3.0 kcal/mol higher than
1
Table 1 differs from that in our earlier paper in that it
b
c
given in the original literature (see text). Reference 20b. Ref-
contains RSE data for five new radicals 8, 12, 21, 22,
and 23, improved values of RSE for radicals 9 and 10
corrected for strain in the precursor, and more recent
ESR splitting constants determined in nonpolar solvents
for most of the radicals listed.
erence 23b. d Reference 23a. e Reference 20d. f Reference 20c. g Ref-
erence 20b. h Kaushal, P.; Pearl, L. H.; Roberts, B. P. J . Chem.
i
j
Soc., Perkin Trans. 2 1990, 1663-1670. Reference 20e. Refer-
ence 20j. This work. Reference 20f. m Reference 20g. Reference
k
l
n
o
2
0i. Korth, H.-G.; Lommes, P.; Sustman, R.; Sylander, L.; Stella,
p
L. New. J . Chem. 1987, 11, 365-375. Paul, H.; Fischer, H. Helv.
R
A plot of a(H ) against RSE for all the appropriate (i.e.,
Chim. Acta 1973, 56, 1575-1588. q Korth, H.-G.; Lommes, P.;
secondary) radicals in Table 1 is shown in Figure 2.
Linear regression affords eq 3 (see Table 4) with coef-
ficients very similar to those given in our earlier paper.
Figure 2 shows that all the species studied lie reasonably
Sicking, W.; Sustmann, R. Chem. Ber. 1985, 118, 4627-4631.
r
s
Reference 20h. Herberg, C. Ph.D. Thesis, University of Freiburg,
t
1
992. Welle, F.; Verevkin, S. P.;Keller, M.; Beckhaus, H.-D.;
R u¨ chardt, C. Chem. Ber. 1994, 127, 697-710.
•
t
close to the line of best fit except 3b ( CHMeOBu ) and
Ta ble 3. BDE(MeXYC-H) Deter m in ed by th e Bor d w ell
Meth od a n d a (Hâ ) for th e Cor r esp on d in g Ra d ica ls
•
Me
4
b ( CHMeNH
2
), the two radicals expected to be some-
what pyramidalized. Each has a value of a(H
G less than those estimated from eq 3.
R
) about 5
a
BDE(C-H) /
-
1
a(HâMe)/G
no.
X
Y
kcal mol
Since in the R u¨ chardt method values of BDE(C-H) are
94.4b
91.2
25c
22.5
21.86
22.8
19.8
19.13
4
45
3
NO2
COEt
H
H
H
d
c
obtained by combination of values of RSE with thermo-
b
e
chemical data for the parent hydrocarbon, a plot of a(H
R
)
5b COPh
8
9
5
0
91
90.5f
e
5
5
-CO2CH2OCO-
against BDE(C-H) should also give a similar linear
correlation. Equation 5, obtained by linear regression
of the appropriate data in Table 1, shows the standard
deviation to be much the same as those for eq 3. Once
again the two nonplanar radicals 3b and 4b are outliers.
b
c
NO2
COPh
COPr
Me
Me
Me
89.8
b
e
a
89
87
80
i
b
g
c
c
6
21
18a
19.9
6.4
pentamethylcyclopentadiene
9-methylfluorene
78.4h
14.56i
We suggested above that a(H
indicator of RSE and BDE(C-H) than a(H
former is less sensitive than the latter to deviations from
planarity of the radical center.4 The plot of a(H
Me) for
â
Me) might be a better
a
The BDE(C-H) data in this table are 3.0 kcal/mol higher than
b
c
R
) because the
given in the original literature. See text. Reference 20e. Ref-
erence 23a. Reference 20d. e Reference 23b. Reference 20k.
d
f
g
Reference 20i. h Reference 20g. This work.
i
â
tertiary radicals against RSE (Figure 3) shows the
validity of this hypothesis. The data point for the oxygen-
•
substituted radical CMe
2
NH
2
4a is much closer than it
•
i
substituted radical CMe
2
OPr 3a lies very close to the
is in the plot of Figure 2. However, as Figure 3 shows,
there are serious discrepancies for the CMe(COMe)
•
line of best fit (eq 4 in Table 4) while that for the amino-
2

1
940 J . Org. Chem., Vol. 63, No. 6, 1998
Ta ble 4. Lin ea r Cor r ela tion s of RSE a n d BDE(CH) w ith ESR P a r a m eter s
Brocks et al.
data collectiona
linear correlations /kcal mol
b
-1
rc
nd
eq
sec
3
4
5
6
7
8
9
0
1
2
3
4
R u¨ chardt
R u¨ chardt
R u¨ chardt
R u¨ chardt
Bordwell
RSE ) 1.65a(HR) - 37.1(( 1.3)e
0.97
0.97
0.96
0.97
0.95
0.95
0.93
0.98
0.96
0.99
0.95
0.93
18
13
18
13
33
8
22
9
8
RSE ) 1.61a(Hâ ) - 36.8(( 1.4)f
t
Me
BDE(C-H) ) 1.61a(HR) + 62.4(( 1.6)
Me
BDE(C-H) ) 1.58a(Hâ ) + 59.4(( 1.3)
BDE(C-H) ) 1.93a(HR) + 57.8(( 2.3)
Me
Bordwell
BDE(C-H) ) 1.45a(Hâ ) + 59.4(( 1.7)
McMillen/Golden
McMillen/Golden
Tsang
Tsang
Pasto
BDE(C-H) ) 1.57a(HR) + 65.2(( 3.2)
Me
1
1
1
1
1
BDE(C-H) ) 1.45a(Hâ ) + 62.8(( 1.4)
BDE(C-H) ) 1.92a(HR) + 57.8(( 1.0)
BDE(C-H) ) 1.21a(Hâ ) + 68.5(( 0.3)g
BDE(C-H) ) 1.68a(HR) + 64.7(( 1.8)
BDE(C-H) ) 1.90a(HR) + 55.1(( 3.0)
Me
5
16
5
Timberlake
a
For citations see text. b a(HR) and a(HâMe) in G. c Correlation coefficient. d Number of data points. e Only valid for secondary radicals.
f
g
Only valid for tertiary radicals. The maximum deviation (0.3 kcal/mol) is smaller than the biggest error estimated by Tsang (0.7 kcal/
mol) in ref 19.
â
F igu r e 3. Plot of RSE against a(H Me) for tertiary radicals
showing the line of eq 4 in Table 4: (4) outliers for the
calculation of eq 4. Compound numbers are shown in Table 1.
â
F igu r e 4. Plot of BDE(C-H) against a(H Me) showing the
line of eq 6: (4) outliers; (s) secondary radicals; (•) tertiary
radicals. Only the tertiary radicals were used to calculate eq
6
in Table 4. Compound numbers are shown in Table 1.
radical 9a and the pentamethylcyclopentadienyl radical
2
8
1 which have a values of a(H
G lower than that expected from eq 4. The case of
â
Me) about 6 G higher and
•
radicals of the type CHMeX. Unfortunately there are
not sufficient data available to enable a reliable correla-
radical 9a is discussed below. The anomalous behavior
of 21 arises from the orbital degeneracy of the cyclopen-
tadienyl radical.25 It is noteworthy that when the
degeneracy is broken by monoalkyl substitution, the
sec
tion to be established. However, the relationship RSE
1.17a(H Me) -30.4 gives satisfactory agreement with
â
all the data available.
The correlation (Figure 4) of a(H Me) against
â
R u¨ chardt’s values of BDE(C-H) (eq 6) is somewhat less
satisfactory than that against RSE (eq 4) although better
)
â
value of a(H Me) is aproximately doubled. Thus the
methylcyclopentadienyl radical has a(H
Me) ) 12.7 G2
6a
â
or 14.8 G.2 The latter value would place 21 very close
6b
than it is for the correlation of a(H
eq 5). The nonplanar radicals 3a , 3b, 4a , and 4b, the
R
) with BDE(C-H)
to the line of eq 4 in Figure 3.
(
A pleasing feature of Figure 3 is that the data for a
wide variety of radicals including captodative radicals,
e.g., 14a , unsaturated species, e.g., 20a , and numerous
substituted alkyl radicals lie close to the line of best fit
over a range of 18 kcal/mol. This engenders confidence
both in the validity of the correlations and in the
reliability of the experimental data.
dicyano and the diacyl radicals 9a and 10a , and the
pentamethylcyclopentadienyl radical 21 all lie off the line
of eq 6. There is considerable divergence between the
values of BDE(C-H) for 9b and 10b obtained from
R u¨ chardt’s data (87.7 and 87.6 kcal/mol, respectively) and
those obtained by Bordwell’s method (95 and 93 kcal/mol,
respectively). The fact that the latter values lie close to
the line of best fit (eq 7) in Figure 5 suggests that the
former require reevaluation. It is noteworthy that for
values of BDE(C-H), unlike those of RSE, the same
correlation (eq 6) applies to both secondary and tertiary
radicals.
•
In the R u¨ chardt method, each secondary radical CHXY
has the same value of RSE as the corresponding tertiary
•
radical CMeXY. However, the values of a(H
â
Me) for the
latter are always smaller than those for the former.
Consequently eq 4 must be modified for secondary
R â
We examined next the correlation of a(H ) and a(H Me)
(
25) For a discussion of the orbital degeneracy of cyclopentadienyl
radical, see ref 26.
26) (a) Kira, M.; Watanabe, M.; Sakurai, H. J . Am. Chem. Soc. 1980,
02, 5202. (b) Barker, P. J .; Davies, A. G.; Tse, M.-W. J . Chem. Soc.,
Perkin Trans. 2 1980, 941-948.
with BDE(C-H) determined by Bordwell’s cyclovoltam-
20
etry/acidity function method. These data were origi-
(
nally calibrated by reference to a scale of 15 values of
1
BDE(C-H) measured in the gas phase.2 However, the
0j

BDE Estimation by ESR Spectroscopy
J . Org. Chem., Vol. 63, No. 6, 1998 1941
F igu r e 5. Plot of Bordwell’s values of BDE(C-H) against
â
F igu r e 6. Plot of BDE(C-H) against a(H Me) showing the
a(H
R
) showing the line of eq 7 in Table 4: (4) outliers.
line of eq 8 in Table 4: (4) outlier. Compound numbers are
Compound numbers are shown in Tables 1 and 2.
shown in Tables 1 and 3.
P r ed ict ion of R SE a n d BDE (C-H) fr om E SR
Da ta . The results described above indicate that ap-
propriate ESR hyperfine splitting constants correlate well
with RSE’s determined by the R u¨ chardt method which
are useful indices of radical stability. Except for non-
gas-phase data were in turn based on values for the
isopropyl and ethyl radical18 that have been shown by
more recent work to be too low.19 Accordingly, for the
present study the Bordwell values of BDE(C-H) have
each been increased by 3.0 kcal/mol.
•
1
2
3
planar radicals, e.g., CR R NR
1, the special features of which are discussed above, they
are predictable from values of a(H ) or a(H Me) by eqs 3
2
, and the radicals 9a or
Tables 2 and 3 show the data for 44 radicals bearing
2
various combinations of the substituents COR, CO
NO , CN, and NR as well as for a variety of unsaturated
radicals. A plot (Figure 5) of a(H ) against BDE(C-H)
2
R, SR,
R
â
2
2
and 4 with the latter generally giving somewhat more
accurate results. The correlations cover a range of about
R
for 35 radicals shows a reasonably good linear correlation
expressed by eq 7, the most serious anomaly being the
cyclopentadienyl radical 59 which, as discussed above,
2
2 kcal/mol and appear to be particularly suitable for
estimating the RSE’s of captodative and benzylic radicals.
For all such radicals, values of RSE predicted from eqs
has a value of a(H
The sulfur-substituted radical, CH
R
) about 6 G less than that expected.
3
or 4 should be accurate to within 2.0 kcal/mol. For
•
2
R
SR 37, has an a(H )
radicals containing an R-methyl substituent, eq 4 is
preferred.
The prediction of the C-H bond dissociation enthalpies
of the parent compounds, which are widely used as
indices of radical stability and reactivity, is more prob-
about 3 G less than that calculated from eq 7, but the
value of BDE(C-H) for this radical is open to doubt
because it was derived from an estimate of the acidity of
the parent compound.
For the nine radicals for which appropriate data are
â
available, there is a good linear correlation of a(H Me)
with Bordwell’s BDE(C-H) expressed by eq 8, the only
serious anomaly being the pentamethylcyclopentadienyl
â
radical 21 which has a(H Me) about 8 kcal/mol lower
than that calculated (Figure 6).
R
lematic. Values of BDE(C-H) calculated from a(H ) by
eqs 5, 7, 9, and 11 are all in reasonable agreement, but
eqs 13 and 14 based respectively on theoretical ap-
proaches give higher and lower values of BDE(C-H).
Values obtained from eq 9 are uniformly ca. 2.0 kcal/mol
above those derived from eq 5, while those calculated
from eq 7 (and eq 11) show a slightly larger divergence
at the top of the range (>22 G) but coincide at about 24.5
G. We consider that either eq 5 based on R u¨ chardt’s
data16 or eq 7 based on Bordwell’s data can confidently
be used to obtain reliable values of BDE(C-H) for planar
or very nearly planar radicals to within a range of 2 kcal/
mol.
An extensive set of data is available in the review by
McMillen and Golden.18 Not surprisingly, in view of the
wide range of sources, there is considerable scatter in a
20
plot of a(H
given by eq 9. A feature of this correlation is that the
values of a(H ) for methyl and some of the primary
R
) against BDE(C-H). The line of best fit is
R
radicals deviate considerably from those calculated. The
same is true of a plot of Tsang’s best values of BDE(C-
Reasonably consistent values of BDE(C-H) can be
1
9
H) against a(H
R
) (eq 11). However, in both cases the
obtained by substitution of the appropriate a(H Me) into
â
linear correlation is much better for the lesser number
Me)
eqs 6, 8, and 10 but eq 12 gives markedly higher values
of data available for plots of BDE(C-H) against a(H
â
at the lower end of the scale. Equations 6 and 8 based
1
6,20
(eqs 10 and 12).
respectively on R u¨ chardt’s and Bordwell’s results
are
Finally, we examined the correlation of a(H
R
) against
derived from the widest range of experimental data, but
values of BDE(C-H) calculated from eq 8 tend to be
about 2.0-2.5 kcal/mol lower than those obtained from
eq 6, probably because of systematic differences in the
values of BDE(C-H) obtained by theoretical methods by
Pasto21 and Timberlake and their co-workers. In both
cases rather poor linear correlations were observed, with
data for radicals expected to be nonplanar lying well
away from the lines of best fit (eqs 13 and 14). The
theoretical values of BDE(C-H) reported by Leroy et
22a
experimental data. Since a(H
â
Me) is less sensitive than
a(H ) to nonplanarity of the radical center, eq 6 is to be
R
preferred over eq 5 if the appropriate ESR hyperfine
coupling constants are available.
2
2b
R
al. did not show a linear correlation with a(H ).

1
942 J . Org. Chem., Vol. 63, No. 6, 1998
Brocks et al.
Ta ble 5. P yr a m id a liza tion Qu otien ts for Ra d ica ls
and tertiary radicals including unsaturated radicals,
benzylic radicals, radicals bearing a single electron-
withdrawing substituent, and a large number of capto-
dative species are either planar or sufficiently close to
planar to allow values of BDE(C-H) and RSE to be
calculated from ESR data by means of eqs 4 and 6.
Radicals bearing fluoro or amino substituents are non-
planar and do not meet the criteria for application of eqs
•
CHMeX
X
a(HR)/G
a(HâMe)/G
Psec
b
b
NH2
NH3
NHCOCH3
NHCONH2
NHEt
NH2Et
NEtCHO
OEt
OCOCH3
OCHO
OP(OEt)2
OPO(OEt)2
OCONH2
OCO2Et
SEt
S(O)Et
S(O2)Et
Br
l4.7
22.7
18.9
16.9
14.6
22.6
18.8
13.9
18.8
l9.6
16.5
18.1
17.6
18.9
16.5
20.2
21.6
20.5
17.3
20.7
26.7
21.8
21.9
20.0
26.9
22.0
21.6
24.0
24.5
22.3
22.6
24.0
23.9
20.3
25.3
27.3
24.7
24.5
1.41
1.18
1.15
1.30
1.37
1.19
1.17
1.55
1.28
1.25
1.35
1.26
1.36
1.26
1.23
1.25
1.26
1.20
l.42
+
a
a
b
b
b
b
a
a
+
a
a
•
1
2
3
a
b
a
a
a
a
a
a
a
a
c
a
a
a
b
a
a
a
a
a
a
a
a
c
a
a
3-6. Although CR R OR are nonplanar and do not
meet the criteria, eq 4 may be applicable. Thus for 3a
eqs 4 and 6 give an RSE of -5.6 kcal/mol and a BDE-
(
(
C-H) of 93.9 kcal/mol in fair agreement with experiment
-5.9 and 93.9 kcal/mol, respectively). However, when
R
the value of a(H ) for 3b is substituted into eq 3, the RSE
obtained (-14.8 kcal/mol) is far too large.
Radicals 29-32 show how the above criteria may be
applied to the estimation of RSE and BDE(C-H). Radi-
cal 29 has a(H
â
Me) ) 15.0 G while 30 has a(H
R
) ) 15.1
2
3b
t
G.
As the pyramidalization quotient P ) 0.99, the
F
values calculated from eqs 3-6 of RSE ) -12.7 kcal/mol
and BDE(C-H) ) 83.1 kcal/mol for compound 29, and
RSE ) -12.2 kcal/mol and BDE(C-H) ) 86.7 kcal/mol
for compound 30 may confidentally be accepted as correct
to within 2 kcal/mol. In accord with this conclusion, the
two values of RSE are in close agreement, while the
difference of 3.6 kcal/mol between the two values of
BDE(C-H) is about what would be expected between a
secondary and a tertiary compound. Conversely, the
a
Reference 23b. b Reference 23b. c Carton, P. M.; Gilbert, B. C.;
Laue, H. A. H.; Norman, R. O. C.; Sealy, R. C. J . Chem. Soc.,
Perkin Trans. 2 1975, 1245.
Although the satisfactory correspondence between the
correlations based on experimental data suggest that
values of a(H ) and a(H Me) should be suitable for the
R â
prediction of BDE(C-H), the plots in Figures 4-6 show
that this is not valid for all radicals. The question arises,
therefore, of how one might assess whether values of
BDE(C-H) obtained in this way are likely to be accurate.
The radicals that deviate most from values predicted by
the appropriate equations are simple saturated hydro-
carbon radicals, radicals that may have unusual strain
or dipolar effects, e.g., 9a , cyclic delocalized radicals, e.g.,
values of a(H
â
Me) ) l6.9 G for 31 and a(H
R
) ) 13.4 G for
23b
t
3
2
give P ) 1.26 well outside the acceptable range.
Clearly, 31 and 32 are significantly pyramidalized and
the values calculated from eqs 3-6 for RSE (-9.6 and
-
15.0 kcal/mol for 31 and 32) and BDE(C-H) (86.1 and
8
4.0 kcal/mol for 31 and 32) must be erroneous.
2
1, and radicals that are nonplanar. The first group
presents no problem as accurate experimental values of
BDE(C-H) are already available. As the experimental
data for some of the radicals in the second group, e.g.,
9
a , are in discord, the accurate prediction of BDE(C-H)
for such species must await agreement on correct values
of BDE(C-H). For the last group (nonplanar radicals),
The data in Table 5 show some interesting trends. For
since values of a(H
R
) are much more sensitive to pyrimi-
Me), we suggest that
example, protonation of R-amino radicals removes the
lone pair from conjugation with the SOMO and virtually
restores planarity. It should be possible to predict values
of RSE and BDE(C-H) for such species. Delocalization
dalization than are values of a(H
â
â R
the ratio a(H Me)/a(H ) be used as an index of deviation
from planarity.27 Table 1 gives values of pyramidaliza-
tion quotients P defined as
of the lone pair by acylation has a similar effect. Thus
t
•
•
•
the 1-(acetamido)ethyl radical CHMeNHCOCH
3
meets
P ) a(H Me( CMeXY))/a(H ( CHXY))
(15)
â
R
the criteria for application of eqs 3-6 as does the
sec
•
•
•
formamido radical CHMeNEtCOH. However, when the
P
) a(H Me( CHXMe))/a(H ( CHXMe)) (16)
â
R
electron-attracting power of the carbonyl group is reduced
•
where X and Y are any substituents. Further values of
by a second donor group as in the radical CHMeNH-
sec
P
obtained from literature data are shown in Table 5.
Our survey of the data presented in the figures and tables
CONH
2
, the nonplanarity is increased.
For oxygenated radicals the situation is less clear-cut.
t
sec
•
•
indicates that if P < 1.05 or if P < 1.25, then values of
RSE and BDE should be obtainable from ESR data. We
recommend that for this purpose eqs 3-6 be used, with
The radicals CHMeOCHO and CHMeSEt just meet the
critera, while the acetate, carbonate, carbamate, phos-
phate, and sulfoxy substituted radicals are slightly
•
preference given to eqs 4 and 6 since a(H
sensitive to pyramidalization of the radical center than
is a(H ). For radicals that meet the criteria, the calcu-
â
Me) is less
outside. The R-bromo radical, CHMeBr, although pos-
sibly somewhat nonplanar, just meets the criteria.
R
lated values of RSE and BDE(C-H) should be accurate
Con clu sion
to within about 2.0 kcal/mol.
t
sec
The above discussion of the correlation of ESR hyper-
fine splitting constants with indices of radical stability
and reactivity indicates that eqs 3 and 5 relating a(H )
R
In general the values of P and P listed in Tables 1
and 5 confirm that a wide range of substituted secondary
to RSE and BDE(C-H) are applicable to a reasonable
variety of substituted planar carbon-centered radicals
(
27) For the first suggested use of this ratio as a measure of radical
pyrimidalization, see ref 4.

BDE Estimation by ESR Spectroscopy
J . Org. Chem., Vol. 63, No. 6, 1998 1943
including captodative radicals. Equation 7 relating a(H
R
)
could also be generated by UV irradiation of 150 µL of ethyl
2
-bromopropionate with 150 µL of Et SiH and 150 µL of DTBP.
3
to BDE(C-H) determined by Bordwell’s method is also
satisfactory over a similar range of radicals. Each of
these equations fails for nonplanar radicals.
3
-Meth yl-2,5-p en ta n ed ion -3-yl (9a ). Generated by UV
irradiation of 30 µL of 3-methyl-2,4-pentanedione with 150 µL
of DTBP and 2.4 mg of trimethylamine-borane in 170 µL of
chlorobenzene: Mod, 1.0 G; T, 195; Pow, 10.1 mW. Simulation
parameters: LS, 52%; LW, 0.43 G; a(3H) ) 21.52 G, a(6H) )
0.1 G.
2,5-P en ta n ed ion -3-yl (9b). Generated from 30 µL of 2,5-
pentanedione by the method used for 9a : Mod, 1.0 G; T, 195;
Pow, 6.4 mW. Simulation parameter: a(1H) ) 18.0 G. The
intensity of the signal decreased rapidly during irradiation.
â
Values of a(H Me) significantly less sensitive to deriva-
tions from planarity of the radical center provide more
reliable indicators of RSE and BDE(C-H) when substi-
tuted into eqs 4, 6, and 8. Although they can be applied
to radicals which deviate slightly from planarity, they
give erroneous results for significantly pyramidalized
radicals. An indication of the degree of planarity and
hence of the applicability of eqs 3-8 can be obtained from
2
-Acetyl-1-h yd r oxy-1-m eth yla llyl (25). Generated by UV
irradiation of 30 µL of 3-methyl-2,4-pentanedione with 80 µL
of DTBP in 90 µL of chlorobenzene: Mod, 0.5 G; T, 195; Pow,
10.1 mW. Simulation parameters: CF, 3327.7 G; LS, 59%;
LW, 0.20 G; a(2H) ) 16.25 G, a(1H) ) 1.46 G, a(6H) ) 3.23 G.
The spectrum also showed the presence of 21% of 9a : CF,
t
sec
pyramidalization quotients (P and P ) derived from the
ratio a(H Me)/a(H ) as shown in eqs 15 and 16. Values
of RSE and BDE(C-H) can be confidently predicted
â
R
sec
t
within a range of 2 kcal/mol when P < 1.25 or P <
l.05. All of the correlations fail for cyclic delocalized
radicals, e.g., cyclopentadienyl, while significant differ-
3
324.1 G.
,1-Dicya n oeth yl (10a ). Generated by UV irradiation of
1
2
9
a saturated solution of methylmalononitrile in 200 µL of 1:1
benzene/DTBP. The two layers so formed were irradiated
inside the cavity of the spectrometer without separation: Mod,
0.5 G; T, 305; SW, 90 G; Pow, 6.4 mW. Simulation param-
eters: LS, 0%; LW, 0.11 G; a(3H) ) 20.64 G, a(2N) ) 2.64 G.
Dicya n om eth yl (10b). Generated by UV irradiation of a
saturated solution of malononitrile in 200 µL of 1:1 benzene/
DTBP: Mod, 0.5 G; T, 280; Pow, 8.1 mW. Simulation param-
eters: LS, 55%; LW, 0.15 G; a(3H) ) 19.22 G, a(2N) ) 2.78 G.
The solution became brown and the signal intensity decreased
very rapidly.
ences between the literature values of BDE(C-H) for
•
some disubstituted species, e.g., CH(COR)
2
, make it
difficult to assess the accuracy of the correlations.
Since extensive compilations of ESR hyperfine splitting
factors are now available, it should be possible to
confidently estimate RSE and BDE(C-H) for a very large
number of radicals to within 2 kcal/mol.
Exp er im en ta l Section
1
,4-Dibu tyl-2,4-diketo-3,6-dim eth ylpiper azin -3-yl (12a).
ESR spectra were obtained on a Bruker ESR 420 spectrom-
eter at an operation frequency of 9.3 GHz. All samples were
recorded in quartz tubes. The samples were degassed by three
cycles of freeze, pump, and thaw and filling with pure nitrogen.
For photochemical reactions, the sample was irradiated with
UV light inside the cavity of the spectrometer by means of a
Generated by UV irradiation of 50 mg of 1,4-dibutyl-3,6-
dimethylpiperazine-2,5-dione with 150 mg of DTBP: Mod, 1.0
G; T, 225; Pow, 9.96 mW. Simulation parameters: LS, 100%;
LW, 1.2 G; a(3H) ) 17.8 G. After 10 min the solution became
red and a second, persistent radical was formed.
1,4-Dibu tyl-2,4-d ik etop ip er a zin -3-yl (12b). Generated
by UV irradiation of 32 mg of 1,4-dibutylpiperazine-2,5-dione
1
000 W high-pressure mercury lamp (Philips). The beam was
passed through a water filter and then focused onto the sample
by a system of two quartz lenses and a mirror. The temper-
ature was controlled by a cooled or heated nitrogen flow
with 220 mg of DTBP in 30 mg of CHCl : Mod, 0.4 G; T, 253;
3
Pow, 4 mW. Simulation parameters: LS, 10%; LW, 0.4 G; a(H)
) 17.1 G, a(2H) ) 5.7 G, a(N) ) 1.2 G, a(2H) ) 1.0 G, a(N) )
0.2 G.
(Bruker). Di-tert-butyl peroxide and halogenated solvents
were filtered over basic alumina. All chemicals were purified
by distillation or recrystallization. The spectra were simulated
with the program NIEHS-WINSIM-EPR version 0.95.28 The
abbreviations are as follows: DTBP, di-tert-butyl peroxide;
Mod, modulation; Pow, microwave power; T, temperature (K);
CF, center field; LS, line shape (% Gaussian); LW, line width.
9-Meth ylflu or en -9-yl (18a ). Generated by heating of 70
mg of 9,9′-bis(9-methylfluorene) in 300 µL of biphenyl ether:
Mod, 0.4 G; T, 475; Pow, 5.1 mW. Simulation parameters: LS,
100%; LW, 0.21 G; a(3H) ) 14.56 G, a(2H) ) 0.83 G, a(2H) )
3.73 G, a(2H) ) 0.81 G, a(2H) ) 3.62 G.
N-(1-E t h oxyca r b on ylet h yl)-N-(m et h yl)a m in om et h yl
(27). Generated by UV irradiation of 50 µL of the ethyl ester
of DL-N,N-dimethylalanine in 150 µL of DTBP: Pow, 8.1 mW;
T, 222. Simulation parameters: LS, 35%; LW, 0.44 G; a(2H)
) 14.30 G, a(3H) ) 4.04 G, a(N) ) 7.00 G, a(1H) ) 1.34 G.
1
-Meth yl-1-(isop r op oxy)eth yl (3a ). Generated by UV
irradiation of 200 µL of diisopropyl ether with 200 µL of
DTBP: Mod, 0.32 G; T, 240; Pow, 6.4 mW. Simulation param-
eters: LS, 34%; LW, 0.35 G; a(6H) ) 19.37 G, a(1H) ) 0.93 G.
1
-Am in o-1-m eth yleth yl (4a ). Generated by UV irradia-
tion of 60 µL of isopropylamine with 170 µL of DTBP: Mod,
.0 G; T, 225; Pow, 6.4 mW. Simulation parameters: LS, 78%;
LW, 0.35 G; a(6H) ) 17.90 G, a(N) ) 3.87 G, a(2H) ) 2.20 G.
-(Eth oxyca r bon yl)eth yl (6b). Generated by UV irradia-
Ack n ow led gm en t. We thank the Deutsche Fors-
chungsgemeinschaft and the Fonds der Chemischen
Industrie for financial support. Part of this work was
carried out while A.L.J .B. was a visitor at the Institut
f u¨ r Organische Chemie and Biochemie der Universit a¨ t
Freiburg. We thank the Alexander von Humboldt Stif-
tung for an award to A.L.J .B.
1
1
tion of 50 µL of ethyl 2-bromopropionate with 50 µL of
hexamethylditin in 150 µL of DTBP: Mod, 0.5 G; T, 293; Pow,
6
.4 mW. Simulation parameters: LS, 100%; LW, 0.11 G; a(3H)
24.68 G, a(2H) ) 1.31 G, a(1H) ) 20.48 G. During the
irradiation a white precipitate accumulated. The same radical
)
J O971940D
(
28) NIEHS WINSIM EPR Version 0.95, Dave Duling, Laboratory
(29) Diez-Barra, E.; de las Hoz, A.; Moreno, A.; S a´ nchez-Verd u´ , P.
J . Chem. Soc., Perkin Trans. 1 1991, 2589-2592.
of Molecular Biophysics NHEIOS, NIH, DHHS.