Chemistry of the t-butoxyl radical: Evidence that most hydrogen abstractions from carbon are entropy-controlled

Article abstract of DOI:10.1021/ja0493493

Absolute rate constants and Arrhenius parameters for hydrogen abstractions (from carbon) by the t-butoxyl radical (^tBuO·) are reported for several hydrocarbons and tertiary amines in solution. Combined with data already in the literature, an an

Full text of DOI:10.1021/ja0493493

Published on Web 05/26/2004
Chemistry of the t-Butoxyl Radical: Evidence that Most
Hydrogen Abstractions from Carbon are Entropy-Controlled
Meghan Finn, Robert Friedline, N. Kamrudin Suleman, Christopher J. Wohl, and
James M. Tanko*
Contribution from the Department of Chemistry, Virginia Polytechnic Institute and
State UniVersity, Blacksburg, Virginia 24061
Received February 5, 2004; E-mail: jtanko@vt.edu
Abstract: Absolute rate constants and Arrhenius parameters for hydrogen abstractions (from carbon) by
the t-butoxyl radical (tBuO‚) are reported for several hydrocarbons and tertiary amines in solution. Combined
with data already in the literature, an analysis of all the available data reveals that most hydrogen abstractions
t
(from carbon) by BuO‚ are entropy controlled (i.e., T∆Sq > ∆Hq, in solution at room temperature). For
substrates with C-H bond dissociation energies (BDEs) > 92 kcal/mol, the activation energy for hydrogen
abstraction decreases with decreasing BDE in accord with the Evans-Polanyi equation, with R ≈ 0.3. For
substrates with C-H BDEs in the range from 79 to 92 kcal/mol, the activation energy does not vary
significantly with C-H BDE. The implications of these results in the context of the use of tBuO‚ as a chemical
model for reactive oxygen-centered radicals is discussed.
Scheme 1
Introduction
In a series of papers starting in 1960,1 Walling and co-workers
reported the results of an intensive study of the mechanism and
kinetics of free radical chlorinations of hydrocarbons by tert-
butyl hypochlorite (tBuOCl) which laid the foundation of our
current understanding of the chemistry of the t-butoxyl radical
(tBuO‚). Initiated by common free radical initiators or light, the
reaction of tBuOCl with hydrocarbons proceeds via a free radical
H-abstraction increased in more polar solvent.3 It was suggested
that the rate of the â-cleavage process increased with solvent
polarity (because of increased polarity in the transition state
associated with the developing carbonyl moiety), but the rate
of H-abstraction was solvent-independent, a hypothesis which
was confirmed nearly three decades later by Lusztyk et al.,4
Tsentalovich et al.,5 and Baciocchi et al.6
t
chain process involving BuO‚ as the hydrogen abstractor
(Scheme 1).
As a hydrogen abstractor, tBuO‚ was found to exhibit modest
selectivity. Competition studies revealed that for alkanes on a
per-hydrogen basis, 3° (40) > 2° (14) > 1° (1). While selectivity
crudely paralleled the strength of the C-H bond, there were
some notable exceptions, e.g., aliphatic C-H bonds were
comparable to benzylic C-H bonds in reactivity, despite an
over 10 kcal/mol difference in bond strength.1
A number of studies also demonstrated that polar effects are
t
important in hydrogen abstractons by BuO‚. For substituted
toluenes, a F-value of -0.4 and a correlation to the σ+
substituent constant was observed (suggesting a slight buildup
of positive charge in the transition state).7 More recent work
by Gleicher8 and Roberts9 provides additional evidence for the
importance of polar effects.
t
It was also found that BuO‚ and other alkoxyl radicals also
undergo unimolecular decomposition (â-cleavage, eq 1). Well
In the intervening 40+ years, a large number of absolute rate
constants and activation parameters for hydrogen abstractions
by tBuO‚ (and also the closely related cumyloxyl radical, C6H5C-
(CH3)2O‚) have been measured, mainly because of technological
advances leading to the development of fast kinetic methods
before the term was coined, this â-cleavage reaction proved to
be a suitable “free radical clock”2 for measuring rates of
(3) Walling, C.; Wagner, P. J. J. Am. Chem. Soc. 1964, 86, 3368-3375.
(4) Avila, D. V.; Brown, C. E.; Ingold, K. U.; Lusztyk, J. J. Am. Chem. Soc.
1993, 115, 466-470.
t
competing bimolecular reactions involving BuO‚. In fact, one
(5) Tsentalovich, Y. P.; Kulik, L. V.; Gritsan, N. P.; Yurkovskaya, A. V. J.
Phys. Chem. A 1998, 102, 7975-7980.
of the few examples of a significant solvent polarity effect on
the rate of a radical reactions was discovered by Walling and
Wagner who found that the relative rate of â-cleavage vs
(6) Baciocchi, E.; Bietti, M.; Salamone, M.; Steenken, S. J. Org. Chem. 2002,
67, 2266-2270.
(7) Russell, G. A. ReactiVity, SelectiVity, and Polar Effects in Hydrogen Atom
Transfer Reactions; Kochi, J. K., Ed.; Wiley: New York, 1973; pp 1-13.
(8) Mahiou, B.; Gleicher, G. J. J. Org. Chem. 1987, 52, 1555-1559.
(9) Roberts, B. P. Chem. Soc. ReV. 1999, 28, 25-35.
(1) Walling, C.; Jacknow, B. B. J. Am. Chem. Soc. 1960, 82, 6108-6112.
(2) Griller, D.; Ingold, K. U. Acc. Chem. Res. 1980, 13, 317-323.
9
7578
J. AM. CHEM. SOC. 2004, 126, 7578-7584
10.1021/ja0493493 CCC: $27.50 © 2004 American Chemical Society

Chemistry of the t-Butoxyl Radical
A R T I C L E S
such as kinetic EPR spectroscopy and laser flash photolysis.10
Indeed, the common belief that the chemistry of tBuO‚ was well-
characterized has led to its use as the prototypical model for
reactive oxygen-centered radicals in a variety of fields. For
Rate constants for hydrogen abstraction are generally expected
to increase with decreasing bond strength. However, this trend
was not observed for tertiary amines. For example, despite a
higher C-H bond strength, triethylamine (R-C-H BDE ) 91.7
kcal/mol)35,36 was found to be four times more reactive than
triallylamine (R-C-H BDE ) 82.6 kcal/mol).35
An analysis of the activation parameters revealed that at room
temperature, the free energy barrier (∆Gq) for the reaction of
tBuO‚ with tertiary amines was dominated more by entropic
than enthalpic factors (i.e., the magnitude of T∆Sq was greater
than ∆Hq). This means that rate constants and selectivities
measured at ambient temperature provide a distorted view of
the intrinsic reactivities of amines in radical reactions and
suggests that (in some cases at least) the chemistry of tBuO‚ is
not sufficiently understood for use as a general chemical model
to mimic the behavior of oxygen-centered radicals.31
t
example, BuO‚ chemistry has been used to test the efficiency
and mechanism of action of antioxidants;11-14 a recent report
points out some of the potential perils of applying this approach
to determine antioxidant mechanism and activity.15 The chem-
istry of tBuO‚ has been used to study mechanisms of oxidative
damage in living (e.g., oxidative damage to DNA,16-18 lipids,19
etc.) and nonliving (e.g., lubricants)20 systems, and also to
predict the oxidative sensitivity of pharmaceuticals.21 In atmo-
t
spheric chemistry, BuO‚ has been used to study mechanisms
of degradation of volatile organic compounds (VOCs) in the
t
stratosphere.22 Finally, BuO‚ has been used in the context of
biomimetic oxidations or chemical model studies to study
mechanisms of enzyme-catalyzed oxidations (e.g., methane
monooxygenase,23 cytochrome P450,24,25 and monoamine oxi-
dase).26 In some model studies of monooxygenase chemistry
using alkyl hydroperoxides and iron complexes, rather than
generating high valent iron-oxo species, alkoxyl radicals such
The scope of the study has been significantly expanded and
activation parameters for twelve new substrates are reported.
These results, taken in conjunction with literature values for
other substrates span C-H bond strengths ranging from 79 to
101 kcal/mol, and provide a comprehensive view of hydrogen
t
as BuO‚ are produced.27-30
t
abstractions from carbon by BuO‚.
In a preliminary communication,we presented evidence
t
Experimental Section
suggesting that there are aspects of BuO‚ chemistry that are
not completely understood. Using laser flash photolysis (LFP),
Materials. All of the solvents and fine chemicals used in this study
were obtained from Aldrich and used as received (except as noted).
The liquid amines and hydrocarbons were vacuum distilled immediately
before use. Diphenylmethanol was sublimed under vacuum immediately
before use. Di-tert-butyl peroxide was purified by passing through a
column of activated alumina.
absolute rate constants and activation parameters were deter-
t
mined for hydrogen abstraction reactions (by BuO‚) from six
tertiary amines (eq 2).31 (Absolute rate constants for hydrogen
Apparatus. Steady-state UV-vis spectra were recorded on a
Hewlett-Packard diode array UV-vis spectrophotometer (HP 8452A).
Laser flash photolysis (LFP) experiments were conducted using an
Applied Photophysic LKS.60 spectrometer using the third harmonic
of a Continuum Surelite I-10 Nd:YAG laser (4-6 ns pulse, 355 nm).
Transient signals were monitored by a Hewlett-Packard Infinium digital
oscilloscope and analyzed with the Applied Photophysics SpectraKinetic
Workstation software package (v. 4.59). Variable temperature experi-
ments were performed with a jacketed cell holder connected to a VWR
Scientific Products (PolyScience) variable temperature circulating bath
(model 1150-A). The cell holder was equipped with a thermocouple
to measure the temperature directly adjacent to the cuvette. Samples
were thermally equilibrated prior to photolysis by placing the cuvettes
in a tray in the circulating bath for at least 10 min. Afterward, the
samples were placed in the spectrometer and equilibrated for an
additional 10 min. (This protocol was checked by placing a thermometer
directly into representative samples and verifying that the internal
temperature was identical to that measured by the thermocouple over
the temperature range of these studies). Concentrations were corrected
for the thermal expansion of the solvent over the temperature range.
Laser Flash Photolysis (LFP). Substrates were prepared in a 2:1
solution of di-t-butyl peroxide:benzene and deoxygenated prior to
photolysis. (Steady-state UV-vis spectra were recorded to verify that
t
abstraction by BuO‚ from tertiary amines had been measured
previously, though no activation parameters were reported.)32-34
(10) Howard, J. A.; Lusztyk, J. Alkoxyl, Carbonyloxyl, Phenoxyl and Related
Radicals; Springer-Verlag: Heidelberg, 1997; Vol. 18: Radical Reaction
Rates in Liquids (Supplement to volume II/13) Subvolume D1; Part 1.
(11) MacFaul, P. A.; Ingold, K. U.; Lusztyk, J. J. Org. Chem. 1996, 61, 1316-
1321.
(12) Valgimigli, L.; Ingold, K. U.; Lusztyk, J. J. Am. Chem. Soc. 1996, 118,
3545-3549.
(13) Jovanovic, S. V.; Steenken, S.; Boone, C. W.; Simic, M. G. J. Am. Chem.
Soc. 1999, 121, 9677-9681.
(14) Sortino, S.; Petralia, S.; Foti, M. C. New J. Chem. 2003, 27, 1563-1567.
(15) Barclay, L. R. C.; Vinqvist, M. R.; Mukai, K.; Goto, H.; Hashimoto, Y.;
Tokunaga, A.; Uno, H. Org. Lett. 2000, 2, 2841-2843.
(16) Adam, W.; Grimm, G. N.; Saha-Moeller, C. R.; DallAÄ cqua, F.; Miolo, G.;
Daniela, V. Chem. Res. Toxicology 1998, 11, 1089-1097.
(17) Adam, W.; Marquardt, S.; Kemmer, D.; Saha-Moeller, C. R.; Schreier, P.
Org. Lett. 2002, 4, 225-228.
(18) Mahler, H.-C.; Schulz, I.; Adam, W.; Grimm, G. N.; Saha-Moeller, C. R.;
Epe, B. Mutation Research 2001, 46, 289-299.
(19) Jones, C. M.; Burkitt, M. J. J. Am. Chem. Soc. 2003, 125, 6946-6954.
(20) Lindsay Smith, J. R.; Nagatomi, E.; Stead, A.; Waddington, D. J.; Be´vie`re,
S. D. J. Chem. Soc., Perkin Trans. 2 2000, 1193-1198.
(21) Karki, S. B.; Treemaneekam, V.; Kaufman, M. J. J. Pharm. Sci. 2000, 89,
1518-1524.
(22) Hudson, A.; Waterman, D.; Alberti, A. J. Chem. Soc., Perkin Trans. 2
1995, 2091-2093.
(23) Choi, S.-Y.; Eaton, P. E.; Hollenberg, P. F.; Liu, K. E.; Lippard, S. J.;
Newcomb, M.; Putt, D. A.; Upadhyaya, S. P.; Xiong, Y. J. Am. Chem.
Soc. 1996, 118, 6547-6555.
(30) MacFaul, P. A.; Arends, I. W. C. E.; Ingold, K. U.; Wayner, D. D. M. J.
Chem. Soc., Perkin Trans. 2 1997, 135-145.
(24) Manchester, J. I.; Dinnocenzo, J. P.; Higgins, L.; Jones, J. P. J. Am. Chem.
Soc. 1997, 119, 5069-5070.
(31) Tanko, J. M.; Friedline, R.; Suleman, N. K.; Castagnoli, N. J. Am. Chem.
Soc. 2001, 123, 5808-5809.
(25) Karki, S. B.; Dinnocenzo, J. P.; Jones, J. P.; Korzekwa, K. R. J. Am. Chem.
Soc. 1995, 117, 3657-3662.
(32) Griller, D.; Howard, J. A.; Marriott, P. R.; Scaiano, J. C. J. Am. Chem.
Soc. 1981, 103, 619-623.
(26) Franot, C.; Mabic, S.; Castagnoli, N. Bioorganic & Medicinal Chemistry
1997, 5, 1519-1529.
(33) Nazran, A. S.; Griller, D. J. Am. Chem. Soc. 1983, 105, 1970-1971.
(34) Dinnocenzo, J. P.; Karki, S. B.; Jones, J. P. J. Am. Chem. Soc. 1993, 115,
7111-7116.
(27) Lindsay Smith, J. R.; Balasubramanian, P. N.; Bruice, T. C. J. Am. Chem.
Soc. 1988, 110, 7411-7418.
(28) Rabion, A.; Chen, S.; Wang, J.; Buchanan, R. M.; Seris, J.-L.; Fish, R. H.
J. Am. Chem. Soc. 1995, 117, 12 356-12 357.
(35) Dombrowski, G. W.; Dinnocenzo, J. P.; Farid, S.; Goodman, J. L.; Gould,
I. R. J. Org. Chem. 1999, 64, 427-431.
(29) Rabion, A.; Buchanan, R. M.; Seris, J.-L.; Fish, H. J. Mol. Catal. A:
Chemical 1997, 116, 43-47.
(36) Wayner, D. D. M.; Clark, K. B.; Rauk, A.; Yu, D.; Armstrong, D. A. J.
Am. Chem. Soc. 1997, 119, 8925-8932.
9
J. AM. CHEM. SOC. VOL. 126, NO. 24, 2004 7579

A R T I C L E S
Finn et al.
Scheme 2
Table 1. Rate Constants and Arrhenius Parameters for Hydrogen
Abstraction by BuO‚ from a Variety of Substrates (Uncertainties in
the last reported digits are in parantheses)
t
log A
T range log(kH) at
E .|
(A in units C−H BDE
(kcal/mol)
a
a
substrate
(°C)
25 °C
(kcal/mol) of M-1 s-1
)
aminesb
tribenzylamine (3)
N,N-dibenzylaniline (2)
DABCO (2)
triallylamine (2)
triethylamine (3)
quinuclidine (2)
5 f 80 7.62 1.34(16) 8.60(10) 89.1(6)c
10 f 80 7.27 2.02(18) 8.75(15) 85(2)c
10 f 70 7.06 2.15(18) 8.63(11) 94(2)d
10 f 80 7.95 2.20(21) 9.56(12) 82.6(8)c
10 f 70 7.99 2.38(49) 9.73(25) 90.7(4)c,e
10 f 70 7.02 2.41(66) 8.78(31) 96(2)d
10 f 80 6.86 2.42(70) 8.62(32) 91(2)f
N-methylpyrrole (1)
N,N-dimethylaniline (2) 10 f 80 8.01 2.54(61) 9.87(29) 92(1)c
hydrocarbons
triphenylmethane (1)
diphenylmethane (2)
allylbenzene (1)
10 f 70 6.31 1.86(36) 7.67(19) 81(1)g
10 f 70 6.53 2.42(53) 8.30(26) 84(1)g
10 f 80 6.51 2.48(31) 8.15(18) 82(2)f
10 f 70 5.91 4.42(78) 9.15(35) 98.7(5)h
di-tert-butyl peroxide was the only species absorbing at the excitation
wavelength). In most LFP experiments, a fixed concentration of
diphenylmethanol (0.100 M) was utilized as a spectroscopic probe,
monitoring the signal buildup at 535 nm (vide infra). Low laser power
(ca. 10-20 mJ) was used in all experiments to eliminate any laser power
dependency to the observed rate constants thereby minimizing the
contributions of radical-radical reactions to the observed rate constant.
Substrate concentrations were varied over a factor of 10 over five
separate experiments (vide infra).
cyclohexane (4)
5.92i 3.11(24)i 8.2(20)i
98.7(5)h
toluene (3)
cyclopentane
t-butylbenzene
10 f 80 5.28 3.46(49) 7.81(24) 90(1)h
-20 f 25 5.93i 3.47(59)i 8.47(45)i 97(2)f
-26 f 10 4.60i 6.14(60)i 9.1(8)i
101(2)f
alcohols
diphenylmethanol (1)
methanol
10 f 70 6.84 2.03(28) 8.33(16) 79(2)f
-35 f 70 6.91j 1.99(35)j 8.37(28)j 79(2)f
-20 f 30 4.72k 5.30(30)k 8.6(2)k
98(1)h
Viscosity Studies. In lieu of a diphenylmethanol probe, only
substrates whose corresponding radicals exhibited a significant
UV-vis absorbance were used and the kinetics followed by monitoring
the buildup of this absorption. The amount of di-tert-butyl peroxide in
solution was decreased to 7.5% to minimize its contribution to the
bulk viscosity of the solution. Substrate concentrations were varied
by over an order of magnitude. Viscosity measurements were per-
formed on each solution using a Oswald viscometer at room temper-
ature. The apparatus constant for the viscometer was measured using
water at 25 °C.
ethers
1,3-dioxolane
-35 f 70 6.90j 3.00(80)j 9.1(7)j
93(2)f
2-methyl-1,3-dioxolane -35 f 70 7.10j 2.09(32)j 8.63(26)j 93(2)f
THF
-35 f 70 6.87j 2.5(10)j 8.7(8)j
-20 f 30 4.99k 5.2(3)k 8.8(2)k
-42 f 20 4.98i 5.21(19)i 8.8(2)i
-20 f 30 4.48k 5.9(3)k 8.8(0.2)k
94(2)f
96(2)f
98(2)f
t-butyl methyl ether
anisole
a Calculated from Arrhenius parameters. b Value in parentheses indicates
number of times study was performed. c Ref 35. d Ref 40. e Ref 36.
f Calculated (B3LYP/6-31G*/CC-PVTZ(-F)) as described herein. g Ref
41. h Ref 42. i Ref 43. j Ref 44. k Ref 45.
Calculations. Density functional theory calculations were performed
using the Titan molecular modeling software (Wavefunction, Inc.,
Irvine, CA 92612 and Schrodinger, Inc., Portland, Oregon, 1998).
Arrhenius parameters for hydrogen abstraction were deter-
mined via nonlinear regression analysis in accordance with the
Arrhenius equation (eq 4);39
Results
t
Rate Constants and Arrhenius Parameters. BuO‚ was
- Ea
RT
generated by flash photolysis of di-tert-butylperoxide (Scheme
2), following the protocol developed by Scaiano et al.37,38
tBuO‚ is spectroscopically invisible at wavelengths > 300 nm,
so the kinetics were followed by monitoring the absorbance of
the products. In most cases, however, the substrates (RH) do
not yield radicals (R‚) which absorb strongly (or at all) at
wavelengths > 300 nm, and a fixed concentration of a spec-
troscopic “probe” was added (typically diphenylmethanol, which
after hydrogen abstraction, yields the diphenylhydroxymethyl
radical which absorbs at 535 nm).37
k ) A exp
(4)
(
)
reported errors are based upon 95% confidence limits of the
parameters. To verify reproducibility, several of the experiments
were repeated over the course of the study; the details are
reported in the Supporting Information. Averaged results are
summarized in Table 1.
In addition to expanding the scope of our earlier study,31 some
of the activation parameters have been revised and are consid-
ered more accurate than the previously reported values. Me-
ticulous purification of the reactants and an extended temperature
range for the Arrhenius studies have led to more reliable and
reproducible values.
Upon the basis of Scheme 2 and eq 3
kobs ) ko + kDPM[Ph2CHOH] + kH[RH]
(3)
Solvent Viscosity Effects on kH. To probe for possible
solvent viscosity effects, the rate constant for hydrogen abstrac-
tion from three tertiary amines (N,N-dimethylaniline, N,N-
dibenzylaniline, and tribenzylamine) were measured in several
n-hydrocarbon solvents and benzene. The results are summarized
in Tables 2 and 3.
for parallel (pseudo) first-order kinetics, the second-order rate
constant for hydrogen abstraction from the substrate (kH) was
obtained from the slope of a plot of the observed rate constant
(kobs) vs [RH] at each temperature. In all cases, at least five
different solutions with RH concentrations varying by at least
1 order of magnitude were studied (at each temperature).
Estimates of C-H Bond Dissociation Energies. When no
experimental values of C-H bond dissociation energies for the
(37) Paul, H.; Small, R. D., Jr.; Scaiano, J. C. J. Am. Chem. Soc. 1978, 100,
4520-4527.
(38) Scaiano, J. C. Nanosecond Laser Flash Photolysis: A Tool for Physical
Organic Chemistry; Moss, R. A., Platz, M. S. and Jones, M., Jr., Ed.; John
Wiley & Sons: Hoboken, 2004, pp 847-871.
(39) Benson, S. W. The Foundations of Chemical Kinetics; McGraw-Hill: New
York, 1960.
9
7580 J. AM. CHEM. SOC. VOL. 126, NO. 24, 2004

Chemistry of the t-Butoxyl Radical
A R T I C L E S
Table 2. Effect of Solvent Viscosity on the Rate Constant for
t
Hydrogen Abstraction from N,N-Dimethylaniline by BuO‚ at 25 °C
solvent
η (cP)
kH (M-1s-1
)
n-C5H12
n-C6H14
n-C8H18
C6H6
n-C14H30
n-C16H34
0.225
0.293
0.518
0.831
1.92
1.2 × 108
1.1 × 108
1.1 × 108
1.1 × 108
9.7 × 107
7.8 × 107
2.9
Table 3. Effect of Solvent Viscosity on the Rate Constant for
Hydrogen Abstraction from N,N-Dibenzylaniline and
t
Tribenzylamine by BuO‚ at 25 °C
kH (M-1s-1
)
kH (M-1s-1
)
kH (M-1s-1
)
in pentane
in benzene
in hexadecane
substrate
(η ) 0.2 cP)
(η ) 0.8 cP)
(η ) 3 cP)
Figure 1. Absolute rate constants for hydrogen abstraction by tBuO‚ (log-
(kH/n), where n is the number of abstractable hydrogens) from a variety of
substrates as a function of C-H bond dissociation energy.
N,N-dibenzylaniline
tribenzylamine
5.5 × 107
4.8 × 107
5.6 × 107
2.2 × 107
1.9 × 107
1.9 × 107
BDEs greater than 92 kcal/mol, log(kH/n) decreases with
increasing bond strength as expected on the basis of the Evans-
Polanyi equation. However, for substrates with C-H BDEs less
than 92 kcal/mol, log(kH/n) seems to level off at a value of about
6.6, independent of C-H BDE.
substrates pertinent to this study were available, they were
estimated using density functional theory46,47 as follows: Ge-
ometry optimizations at the B3LYP/6-31G* level followed by
single point energy calculations (B3LYP/CC-PVTZ(-F)), were
performed on the substrate and its corresponding free radical
to obtain the difference in their energy (∆E). Using 17
compounds with known C-H BDEs (ranging from 81 to 132
kcal/mol), a plot of BDE vs ∆E was constructed and analyzed
by linear least squares regression analysis. Unknown C-H
BDEs were estimated by calculating ∆E using DFT as described
above, and using the results of the regression analysis (see
Supporting Information). Literature and calculated values of
pertinent C-H BDEs are also summarized in Table 1.
At first glance, these results might be explained on the basis
of the reactivity-selectivity principle. The strength of the O-H
bond in tBuOH is 105 kcal/mol,42 which means that all of these
reactions are exothermic by 3-25 kcal/mol, and because of its
high reactivity, tBuO‚ is expected to exhibit low selectivity. In
the context of the Evans-Polanyi relationship, a low value of
R is expected, consistent with an early, reactant-like transition
state. On this basis, despite having C-H bond which is weaker
by 8 kcal/mol, triallylamine and triethylamine react at nearly
Discussion
t
the same rate with BuO‚. Similarly, on a per hydrogen basis,
Rate constants for hydrogen abstraction are expected to
increase with decreasing bond strength. The classic method for
conceptualizing this is based upon the Evans-Polanyi relation-
ship which predicts a linear relationship between the log of the
rate constant for hydrogen abstraction and the enthalpy of the
reaction: log(kH) ) R∆H° + constant. ∆H° is directly related
to the strength of the breaking bond; for H-abstractions by
tBuO‚, ∆H° ) BDE(C-H) - BDE(OH). The proportionality
constant R is believed to be a measure of transition state
location: A value of R < 0.5 implies an early, more reactant-
like transition state, whereas R > 0.5 implies a late, more
product-like transition state.7
Using values for kH at 25 °C (Table 1), a plot of log(kH/n) vs
BDE (where n is the number of abstractable hydrogens) is
presented in Figure 1. Upon the basis of the observed scatter,
no simple relationship between log(kH/n) and C-H BDE in the
context of the Evans-Polanyi relationship is evident. Through
the scatter, however, there appears to be curvature in the plot,
with two distinct regions discernible: For substrates with C-H
the rate constant for reaction of cyclohexane and toluene are
the same, despite the fact that the C-H bond in toluene is 10
kcal/mol weaker. At the extreme, the high reactivity/low
selectivity argument would suggest that every encounter of
tBuO‚ would lead to reaction (i.e., the onset of diffusion control),
and this might explain why log(kH/n) does not vary significantly
with structure for substrates with low C-H BDEs. However,
the observed bimolecular rate constants are lower than values
typically associated with a diffusion-controlled reactions in
solution (109-1010 M-1s-1).48
To understand the reason for this apparent breakdown in
normal activation/driving force relationships, it is critical to go
beyond the rate constants and examine the activation parameters
for hydrogen abstraction by tBuO‚. Recalling the relationships
between log(A) and Ea of the Arrhenius equation and the
enthalpy and entropy of activation of activated complex theory
(eqs 5 and 6),39
∆H q ) Ea - RT
(5)
(6)
(40) Laleve´e, J.; Allonas, X.; Fouassier, J. P. J. Am. Chem. Soc. 2002, 124,
9613-9621.
h
∆S q ) R ln
+ 2.303log A - 1
( (
)
)
kBT
(41) McMillen, D. F.; Golden, D. M. Annu. ReV. Phys. Chem. 1982, 33, 493-
532.
(42) Tsang, W. Heats of Formation of Organic Free Radicals by Kinetic
Methods; Simo˜es, J. A. M., Greenberg, A. and Liebman, J., Ed.; Chapman
& Hall: New York, 1996; pp 22-58.
in Figure 2 a plot of -T∆Sq vs ∆Hq is presented, constructed
from these results and data already reported in the literature;
(43) Weber, M.; Fischer, H. J. Am. Chem. Soc. 1999, 121, 7381-7388.
(44) Malatesta, V.; Scaiano, J. C. J. Org. Chem. 1982, 47, 1455-1459.
(45) Wong, S. K. J. Am. Chem. Soc. 1979, 101, 1235-1239.
(46) Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5651.
(47) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B. 1988, 37, 785-789.
(48) Tanko, J. M.; Suleman, N. K. SolVent Effects in the Reactions of Neutral
Free Radicals; Simo˜es, J. A. M., Greenberg, A., Liebman, J., Eds.; Chapman
& Hall: New York, 1996; pp 224-293.
9
J. AM. CHEM. SOC. VOL. 126, NO. 24, 2004 7581

A R T I C L E S
Finn et al.
t
Figure 3. Activation energy for hydrogen abstraction by BuO‚ vs C-H
BDE.
t
Figure 2. Plot of -T∆Sq vs ∆Hq for hydrogen abstractions by BuO‚ at
room temperature.
the diagonal line corresponds to -T∆Sq ) ∆Hq (at 298 K). It
should be noted that the derived ∆Hq and ∆Sq refer to a
reference state where one mole of reactants, each at 1 M
concentration, are converted to one mole of the transition state,
also at 1 M concentration.
t
A critical feature about BuO‚ chemistry emerges from this
analysis: At room temperature, most hydrogen abstractions
t
by BuO‚ in solution are entropy-controlled. What this means
is that the free energy barrier (∆Gq ) ∆Hq - T∆Sq) for
hydrogen abstraction is governed more by the entropy rather
than the enthalpy of activation. This point is illustrated graph-
ically in Figure 2, which clearly shows for most substrates at
room temperature, -T∆Sq > ∆Hq (i.e., most of the points in
Figure 2 fall aboVe the line corresponding to -T∆Sq ) ∆Hq).
This means alkoxyl radicals are so reactive that rates of
hydrogen abstractions are governed more by issues of orienta-
tion, trajectory, accessibility, etc., than by the strength of the
C-H bond. Entropy-controlled reactions are not common in
organic chemistry and do not follow “normal” structure/reac-
tivity and/or linear free energy relationships. Examples of re-
ported entropy-controlled reactions include additions of carbenes
to multiple bonds49,50 and radical-radical recombination reac-
tions.51 Some of these reactions are characterized by very low
activation energies, or in some cases, even negative activation
energies and/or curved Arrhenius plots, low A-factors, and rate
constants below the diffusion-controlled limit. For all substrates
examined in this study, “normal” Arrhenius behavior was
observed.
Does the activation energy (or ∆Hq) vary sensibly with the
strength of the C-H bond? A plot of activation energies vs
BDE (Figure 3) reveals two distinct regions: For substrates with
C-H BDEs greater than about 92 kcal/mol, the activation
energy decreases as the C-H bond becomes weaker (as is
normally observed for hydrogen abstraction processes). From
this region of the plot, an R value of about 0.3 is obtained,
suggesting an early transition state reminiscent of other high
reactivity/low selectivity radicals such as Cl‚.7 (However, unlike
Figure 4. Solvent viscosity effectshydrogen abstraction from N,N-
dimethylaniline by BuO‚ compared to a bona fide diffusion controlled
reaction (data from Table 2 and ref 53).
t
Cl‚,52 the rate constants are well below the diffusion-controlled
limit). For substrates with C-H BDEs less than 92 kcal/mol,
1
the activation energy levels off to a value of about 2 (( /2)
kcal/mol and does not vary significantly with C-H BDE
(R ) 0).
An activation energy of ca. 2 kcal/mol is very close to the
activation energy for viscous flow for relatively low viscosity
solvents such as benzene, and the possibility that the level region
of Figure 3 might be attributable to the onset of diffusion control
was addressed. This was accomplished by examining the effect
of solvent viscosity on kH for substrates with BDEs less than
92 kcal/mol and with low Ea’s. Upon the basis of the Stokes-
Einstein and von Smoluchowski equations, it is expected that
the diffusion controlled rate constant (kdiff) will vary linearly
with the inverse of solvent viscosity (at constant temperature).48
In Figure 4, the effect of solvent viscosity for the reaction of
tBuO‚ with N,N-dimethylaniline (C-H BDE ) 91.7 kcal/mol;
(49) Moss, R. A.; Lawrynowicz, W.; Turro, N. J.; Gould, I. R.; Cha, Y. J. Am.
Chem. Soc. 1986, 108, 7028-7032.
(50) Houk, K. N.; Rondan, N. G. J. Am. Chem. Soc. 1984, 106, 4293-4294.
(51) Sobek, J.; Martschke, R.; Fischer, H. J. Am. Chem. Soc. 2001, 123, 2849-
2857.
(52) Bunce, N. J.; Ingold, K. U.; Landers, J. P.; Lusztyk, J.; Scaiano, J. C. J.
Am. Chem. Soc. 1985, 107, 5464-5472.
9
7582 J. AM. CHEM. SOC. VOL. 126, NO. 24, 2004

Chemistry of the t-Butoxyl Radical
A R T I C L E S
t
Ea ) 2.54 kcal/mol) is compared to the effect of solvent
viscosity on a bona fide diffusion controlled reactions
combination/disproportionation of the tert-butyl radical (using
data of Schuh and Fischer).53 This figure clearly demonstrates
that there is no significant effect of solvent viscosity on the
rate constant for hydrogen abstraction from N,N-dimethylaniline
transition state for the reaction of BuO‚ and Ph2CH, only one
of the two phenyl groups is oriented properly to stabilize the
developing radical center; the other is twisted out of conjugation.
Similarly for Ph3CH + tBuO‚, only one of three phenyl groups
is oriented properly. Hence, the fact that the activation energies
for PhCH3, Ph2CH2, and Ph3CH are so similar may be
attributable that only one phenyl group is available for stabilizing
the transition state in each instance.
t
by BuO‚. For other substrates in the “level” region of the Ea
vs C-H BDE plot such as N,N-dibenzylaniline (C-H BDE )
89.1 kcal/mol; Ea ) 2.01 kcal/mol) and tribenzylamine (C-H
BDE ) 85.4 kcal/mol; Ea ) 1.34 kcal/mol), a 1 order of
magnitude variation in solvent viscosity had no discernible effect
on kH (Table 3). Consequently, it appears that the leveling off
observed in the Ea vs C-H BDE plot cannot be attributed to
the onset of diffusion control.
2. For the reaction tBuO‚ + tertiary amines, stereoelectronic
considerations are important32 and interaction between the
developing radical center and nitrogen lone pair in the transition
state may be more important than interactions with other, radical
stabilizing substituents (eq 7). However, it should be noted that
There are several possible explanations for the observation
that the activation energies are independent of BDE (for
substrates with C-H BDE < 92 kcal/mol), including the
possibility that this is an experimental artifact. There are errors
associated with the measurement of both Ea and the C-H BDE,
and a fair amount of scatter in the data. Could these errors
obscure what is actually a linear relation between these two
quantities?
With regard to errors in the reported activation energies, there
has been some discussion in the literature regarding what is
often referred to as a compensation effect in regression analysis
(i.e., within limits of uncertainty, the data can be fit equally
well with a lower Ea/log(A) or higher Ea/log(A).54 At the present
time, there appears to be no clear consensus as to how to report
the error in Ea or log(A) to account for this compensation effect;
standard practice is to report standard errors or 95% confidence
limits derived from the regression analysis. Also, there has been
discussion of the use of the linearized vs nonlinearized forms
of the Arrhenius equation, and the issue of proper weighting of
the data.54-56
In these experiments, a temperature range of ca. 10-80 °C
was used to avoid freezing of the solution, or thermal decom-
position of the tBuO‚ precursor, di-tert-butyl peroxide). For our
data in Table 1, 95% confidence limits are provided for Ea and
log(A) based upon the regression statistics. Under the assumption
that the error in k was independent of temperature, these values
are derived from fitting of the nonlinear equation. However, it
should be noted that within reported error limits, the results are
identical using the linearized form of the Arrhenius equation.
Finally, in addition to spot-checking our technique by repeating
Arrhenius studies previously reported in the literature, several
of the new experiments were performed and repeated several
times in their entirety over the course of this study. Accordingly,
we are confident that the values of the Arrhenius parameters
reported in this investigation and in the literature are accurate
and reproducible.
for quinuclidine and DABCO, the nitrogen lone pair cannot
provide this stabilization, yet the activation energies are only
slightly greater than that of the other amines which do not
possess this geometric constraint.
3. These hydrogen abstractions may be so exothermic that
there is not a significant barrier associated with the bond making/
bond breaking process. The nominal activation energies may
result from residual steric interactions between the bulky tert-
butyl group and the substrate.
4. Somewhat related to #3, Houk has examined the flexibility
of the transition state for the reaction of hydroxyl radical and
methane (HF/3-21G and MP2/6-31G**) and found that length-
ening the C-O bond in the transition state by 0.1 Å over the
optimum length increases the activation energy by 1.2 kcal/
mol.57 For substrates with weak C-H bonds, although one
would expect that the Ea would decrease with increasing bond
t
strength, this trend may be interrupted because BuO‚ cannot
get close enough to the substrate in the transition state to attain
the optimal bond length.
The experimental results do not allow us to determine which
(if any) of the possibilities is most likely and in future work,
this will be addressed by current theories pertaining to hydrogen
abstractions reactions (e.g., the Zavitsas nonparametric model
for calculating activation energies of hydrogen abstraction
reactions,58,59 Roberts’ extended Evans-Polanyi relationship,60
and recent approaches offered by Mayer.)61,62 Ab initio calcula-
tions of transition state structures and energetics will also be
performed.
The final issue to consider pertaining to activation parameters
for hydrogen abstraction by tBuO‚ is the magnitude of log(A),
which via eq 6, is directly related to the entropy of activation.
The statement that most hydrogen abstractions (from carbon)
by tBuO‚ are entropy-controlled does not necessarily mean that
the entropy requirements in the transitions states for these
reactions are unusually stringent. For the substrates listed in
In terms of a “chemical” explanation for these results, we
offer the following hypotheses:
1. In some instances, steric interactions attributable to the
t
bulky BuO‚ may force other, radical stabilizing substituents
on the substrate out of conjugation with the developing radical
center. For example, in the calculated (B3LYP/6-31G**)
(57) Dorigo, A. E.; Houk, K. N. J. Org. Chem. 1988, 53, 1650-1664.
(58) Zavitsas, A. A.; Melikian, A. A. J. Am. Chem. Soc. 1975, 97, 2759-2763.
(59) Zavitsas, A. A.; Chatgilialoglu, C. J. Am. Chem. Soc. 1995, 117, 10 645-
10 654.
(53) Schuh, H.; Fischer, H. Int. J. Chem. Kinet. 1976, 8, 341-356.
(54) He´berger, K.; Keme´ny, S.; Vido´czy, T. Int. J. Chem. Kinet. 1987, 19, 171-
181.
(55) Calverley, E. M. AIChE J. 1993, 39, 725-726.
(56) Klicka, R.; Kuba´cek, L. Chemom. Intell. Lab. Syst. 1997, 39, 69-75.
(60) Roberts, B. P.; Steel, A. J. J. Chem. Soc., Perkin Trans. 2 1994, 2155-
2162.
(61) Mayer, J. M. Acc. Chem. Res. 1998, 31, 441-450.
(62) Bryant, J. R.; Mayer, J. M. J. Am. Chem. Soc. 2003, 125, 10 351-10 361.
9
J. AM. CHEM. SOC. VOL. 126, NO. 24, 2004 7583

A R T I C L E S
Finn et al.
stated: (1) At room temperature, most hydrogen abstractions
(from carbon) by tBuO‚ in solution are entropy-controlled; the
free energy barrier is governed more by T∆Sq than by ∆Hq. (2)
For substrates with C-H BDE > 92 kcal/mol, normal reactivity
trends are observedsthe activation energy decreases with
decreasing C-H bond strength. (3) For substrates with C-H
BDE < 92 kcal/mol, the activation energy levels off to a value
of approximately 2 kcal/mol. Solvent viscosity studies demon-
strate that this leveling is not attributable to the onset of diffusion
control.
t
As noted in the Introduction, the chemistry of BuO‚ has
frequently been used as a chemical model for reactive oxygen-
centered radicals in a variety of contexts, and some comments
regarding the viability of this approach are appropriate. For
t
reactive substrates, hydrogen abstractions by BuO‚ are domi-
t
nated by entropy considerations. This means that any intrinsic
reactivity trends associate with a substrate, or class of substrates,
are masked. It appears that much of the behavior associated
Figure 5. Variation of log(A/n) for hydrogen abstractions by BuO‚ as a
function of C-H BDE (n is the number of abstractable hydrogens).
Table 1, ∆Sq ranges from -8 to -19 eu. These values are
smaller than bimolecular reactions with a high degree of order
in the transition state such as the Diels-Alder reaction. For
example, for the dimerization of cyclopentadiene63 and reaction
of maleic anhydride with 9,10-dimethylanthracene,64 ∆Sq’s are
-29 and -39 eu, respectively.
At this time, it is not clear whether log(A) varies in any sort
of systematic way with structure. A plot of log(A/n) vs C-H
BDE (Figure 5, where n is the number of abstractable
hydrogens) shows that for most substrates, log(A/n) hovers
around 8.0, and it appears that tertiary amines consistently have
a slightly higher log(A/n) than aromatic hydrocarbons. Consid-
ered in light of the experimental error associated with these
values and scatter in Figure 5, it is inappropriate to draw any
meaningful conclusions at this time. (We plan on addressing
this matter in the future via the ab initio calculations mentioned
earlier).
t
with BuO‚ is unique, likely arising from numerous issues
pertaining to steric bulk. Hence, tBuO‚ may not be representative
of other (smaller) alkoxyl radical, and other reactive oxygen-
centered species in general.
Acknowledgment. Financial support from the National Sci-
ence Foundation (CHE-0108907) is acknowledged and appreci-
ated. We also thank Profs. Neal Castagnoli, James Espenson,
and Andreas Zavitsas for stimulating conversations.
Supporting Information Available: Tables summarizing
absolute rate constants as a function of temperature and pertinent
Arrhenius plots for all substrates used in this study, a summary
of Arrhenius parameters obtained in individual trials, results of
B3LYP calculations to estimate unknown C-H bond dissocia-
tion energies of several substrates, and calculated transition states
t
for the reaction of BuO‚ with toluene and triphenylmethane.
This material is available free of charge via the Internet at
http://pubs.acs.org.
Conclusions
Although the results presented in this paper raise many new
questions regarding the chemistry of alkoxyl radicals that can
only be addressed by additional work (theory and experiment),
there are a number of significant conclusions that can be
JA0493493
(63) Walling, C.; Schugar, H. J. J. Am. Chem. Soc. 1963, 85, 607-612.
(64) Andrews, L. J.; Keefer, R. M. J. Am. Chem. Soc. 1955, 77, 6284-6289.
9
7584 J. AM. CHEM. SOC. VOL. 126, NO. 24, 2004