Abnormal solvent effects on hydrogen atom abstraction. 2. Resolution of the curcumin antioxidant controversy. The role of sequential proton loss electron transfer

Article abstract of DOI:10.1021/jo049254j

The rates of reaction of 1,1-diphenyl-2-picrylhydrazyl (dpph^.) radicals with curcumin (CU, 1,7-bis(4-hydroxy-3-methoxyphenyl)-1,6-heptadiene-3, 5-dione), dehydrozingerone (DHZ, "half-curcumin"), and isoeugenol (IE) have been measured in methanol and ethanol and in two non-hydroxylic solvents, dioxane and ethyl acetate, which have about the same hydrogen-bond-accepting abilities as the alcohols. The reactions of all three substrates are orders of magnitude faster in the alcohols, but these high rates can be suppressed to values essentially equal to those in the two non-hydroxylic solvents by the addition of acetic acid. The fast reactions in alcohols are attributed to the reaction of dpph^. with the CU, DHZ, and IE anions (see J. Org. Chem. 2003, 68, 3433), a process which we herein name sequential proton loss electron transfer (SPLET). The most acidic group in CU is the central keto-enol moiety. Following CU's ionization to a monoanion, ET from the [-(O)CCHC(O)-]^- moiety to dpph^. yields the neutral [-(O)CCHC(O)-]^. radical moiety which will be strongly electron withdrawing. Consequently, a phenolic proton is quickly lost into the alcohol solvent. The phenoxide anion so formed undergoes charge migration to produce a neutral phenoxyl radical and the keto-enol anion, i.e., the same product as would be formed by a hydrogen atom transfer (HAT) from the phenolic group of the CU monoanion. The SPLET process cannot occur in a nonionizing solvent. The controversy as to whether the central keto-enol moiety or the peripheral phenolic hydroxyl groups of CU are involved in its radical trapping (antioxidant) activity is therefore resolved. In ionizing solvents, electron-deficient radicals will react with CU by a rapid SPLET process but in nonionizing solvents, or in the presence of acid, they will react by a slower HAT process involving one of the phenolic hydroxyl groups.

Full text of DOI:10.1021/jo049254j

Abn or m a l Solven t Effects on Hyd r ogen Atom Abstr a ction . 2.
Resolu tion of th e Cu r cu m in An tioxid a n t Con tr over sy. Th e Role of
Sequ en tia l P r oton Loss Electr on Tr a n sfer
Grzegorz Litwinienko*,† and K. U. Ingold
National Research Council, Ottawa, Ontario, Canada K1A 0R6, and Warsaw University,
Department of Chemistry, Pasteura 1, 02-093 Warsaw, Poland
litwin@chem.uw.edu.pl
Received May 3, 2004
The rates of reaction of 1,1-diphenyl-2-picrylhydrazyl (d p p h •) radicals with curcumin
(CU, 1,7-bis(4-hydroxy-3-methoxyphenyl)-1,6-heptadiene-3,5-dione), dehydrozingerone (DHZ, “half-
curcumin”), and isoeugenol (IE) have been measured in methanol and ethanol and in two non-
hydroxylic solvents, dioxane and ethyl acetate, which have about the same hydrogen-bond-accepting
abilities as the alcohols. The reactions of all three substrates are orders of magnitude faster in the
alcohols, but these high rates can be suppressed to values essentially equal to those in the two
non-hydroxylic solvents by the addition of acetic acid. The fast reactions in alcohols are attributed
to the reaction of d p p h • with the CU, DHZ, and IE anions (see J . Org. Chem. 2003, 68, 3433), a
process which we herein name sequential proton loss electron transfer (SPLET). The most acidic
group in CU is the central keto-enol moiety. Following CU’s ionization to a monoanion, ET from
the [-(O)CCHC(O)-]- moiety to d p p h • yields the neutral [-(O)CCHC(O)-]• radical moiety which
will be strongly electron withdrawing. Consequently, a phenolic proton is quickly lost into the alcohol
solvent. The phenoxide anion so formed undergoes charge migration to produce a neutral phenoxyl
radical and the keto-enol anion, i.e., the same product as would be formed by a hydrogen atom
transfer (HAT) from the phenolic group of the CU monoanion. The SPLET process cannot occur in
a nonionizing solvent. The controversy as to whether the central keto-enol moiety or the peripheral
phenolic hydroxyl groups of CU are involved in its radical trapping (antioxidant) activity is therefore
resolved. In ionizing solvents, electron-deficient radicals will react with CU by a rapid SPLET
process but in nonionizing solvents, or in the presence of acid, they will react by a slower HAT
process involving one of the phenolic hydroxyl groups.
CHART 1. Str u ctu r es a n d Abbr evia tion s of th e
Ma in Com p ou n d s Stu d ied : Cu r cu m in (CU) in th e
r,γ-Dik eto a n d Keto-En ol F or m s (Str u ctu r es 1a
a n d 1b, Resp ectively), Deh yd r ozin ger on e (DHZ),
a n d Isoeu gen ol (IE)
In tr od u ction
Curcumin (CU), 1,7-bis(4-hydroxy-3-methoxyphenyl)-
1,6-heptadiene-3,5-dione, 1a , the yellow pigment of tur-
meric and curry, exists mainly in the keto-enol form,
1b (see Chart 1), which is favored by a strong intramo-
lecular hydrogen bond.1,2
Many health benefits have been claimed for CU,3 and
these have generally been ascribed to its radical-trapping
antioxidant properties.4 Since CU is a (bis)phenol, its
reported ability to trap lipid peroxyl5,6 and 1,1-diphenyl-
2-picrylhydrazyl7 (d p p h •) radicals by donating one of its
phenolic H-atoms is consistent with the known mecha-
* To whom correspondence should be addressed. Phone: 48-22-
8220211 ext. 335. Fax: 48-22-8225996.
† Warsaw University.
(1) Tønnesen, H. H. In Phenolic compounds in food and their effects
on health. I. Analysis, Occurrence, and Chemistry; Ho, C.-T., Lee, C.
Y., Huang, M. T., Eds.; ACS Symposium Series 506; American
Chemical Society: Washington, DC, 1992; pp 143-153.
(2) Roughley, P. J .; Whiting, D. A. J . Chem. Soc., Perkin Trans. 1
1973, 2379-2388. Pedersen, U.; Rasmussen, P. B.; Lawesson, S.-O.
Liebigs Ann. Chem. 1985, 1557-1569.
(3) E.g., reduction in blood cholesterol, inhibition of low-density
lipoprotein peroxidation, platelet aggregation, HIV replication, cataract
formation, etc. Reviews: Leu, T.-H.; Maa, M.-C. Curr. Med. Chem.:
Anti-Cancer Agents 2002, 2, 357-370. Aggarwal, B. B.; Kumar, A.;
Bharti, A. C. Anticancer Res. 2003, 23, 363-398.
nism by which other phenols trap peroxyls,8 alkoxyls,9-11
and d p p h •,9,11,12 reaction 1.
X• + ArOH f
XH + ArO• (X• ) ROO•, RO•, d p p h •, etc.) (1)
10.1021/jo049254j CCC: $27.50 © 2004 American Chemical Society
Published on Web 08/05/2004
5888
J . Org. Chem. 2004, 69, 5888-5896

Abnormal Solvent Effects on Hydrogen Atom Abstraction
However, since curcumin’s phenolic hydrogen atoms
are intramolecularly H-bonded to the adjacent methoxy
groups, it is expected to be a relatively poor hydrogen
atom transfer (HAT) agent, as is illustrated by the rate
CH2 group and from the enolic OH group to form a
carbon-centered radical which reacted rapidly with di-
oxygen, a process which eventually leads to C-C bond
cleavage, reactions 3 and 4.
constants for reaction of d p p h • in alkane solvents at 298
K with 2-methoxyphenol (0.7 M-1 s-1
)
compared with
12
4-methoxyphenol (240 M-1 s-1).11 Thus, a statement that
CU is a “superb antioxidant”13 seems improbable, and
at first sight, it is surprising that CU’s radical trapping
properties have received such attention.14 There is no
doubt that much of this attention can be attributed to
suggestions that the R,γ-dicarbonyl moiety (both as the
diketone, 1a , and as the keto-enol, 1b) is involved in
radical trapping.13,15-18 However, “radical trapping” does
not, in and of itself, make an antioxidant. For a compound
to be a radical-trapping antioxidant it is essential that
the antioxidant-derived radical does not react with di-
oxygen as this would continue the autoxidation chain.
For this reason, among others, phenols are radical-
trapping antioxidants; i.e., reaction 2 does not occur.
Thus, the R,γ-diketone moiety may “trap” peroxyl
radicals, but this does not make CU a radical-trapping
antioxidant. Also in 1996, Sreejayan and Rao7a reported
that a diacetylated CU in which both phenol OH groups
had been converted to acetyl groups did not react with
d p p h • in ethanol. Similarly, Priyadarsini et al.7b,c have
reported recently that a methylated CU in which both
phenolic OH groups had been converted to OCH3 groups
reacted with d p p h • 1800 times more slowly than CU.
Despite this background, in 1999, J ovanovic et al.13
claimed that CU was a “superb H-atom donor.” The
diketo form was (possibly incorrectly)19-21 assumed to be
the dominant form in their experiments and was “an
extraordinarily potent H-atom donor...due to delocaliza-
tion of the unpaired electron on the adjacent oxygens”
(shown as in reaction 5). Certainly, the reported rate
constants for H-atom abstraction from the central CH2
group of CU by methyl radicals (3.5 × 109 M-1 s-1 in 40%
aqueous DMSO at pH 5) and tert-butoxyl radicals (7.5 ×
109 M-1 s-1 in acetonitrile) are extraordinarily high.
Indeed, they would appear to be impossibly high for
abstraction of H from any C-H moiety.22
ArO• + O2 N
(2)
In this connection, it is important to note that in 1996
Sugiyama et al.17 showed that the radicals derived from
R,γ-dicarbonyl moieties do react with dioxygen. Specifi-
cally, these workers showed that when dimethoxytet-
rahydrocurcumin (CU modified by reduction of the two
vinyl (styrene) groups and by methylation of the two
phenolic OH groups) was oxidized with peroxyl radicals
in oxygen-saturated acetonitrile it underwent C-C bond
cleavage at the -(O)CCH2C(O)- moiety. Bond cleavage
was attributed to H-atom abstraction from the central
(4) In iron-catalyzed lipid peroxidation the antioxidant activity of
CU and its acetylated derivatives has been ascribed to iron chelation
by the R,γ-diketone moiety; e.g., see: Sreejayan, N.; Rao, M. N. A. J .
Pharm. Pharmacol. 1994, 46, 1013-1016. See also: Began, G.;
Sudharshan, E.; Udaya Sankar, K.; Appu Rao, A. G. J . Agric. Food
Chem. 1999, 47, 4992-4997.
(5) Priyadarsini, K. I. Free Rad. Biol. Med. 1997, 23, 838-843.
(6) Barclay, L. R. C.; Vinqvist, M. R.; Mukai, K.; Goto, H.; Hash-
imoto, Y.; Tokunaga, A.; Uno, H. Org. Lett. 2000, 2, 2841-2843.
(7) (a) Sreejayan, N.; Rao, M. N. A. Arzneim.-Forsch./ Drug Res.
1996, 46, 169-171. (b) Priyadarsini, K. I.; Maity, D. K.; Naik, G. H.;
Kumar, M. S.; Unnikrishan, M. K.; Satav, J . G.; Mohan, H. Free Rad.
Biol. Med. 2003, 35, 475-484. (c) In ref 7b the text states that the
solvent used in the d p p h •/CU kinetic study was acetonitrile, but the
rate constant given in Table 1 is said to have been measured in
methanol!
J ovanovic et al.’s13 conclusions were firmly rejected in
2000 by Barclay et al.6 These workers examined the
actual antioxidant activities of CU, three curcumin
analogues with no phenolic hydroxyl groups and three
2-methoxy-4-alkylphenols by measuring their abilities to
inhibit the autoxidation of styrene and methyl linoleate
in chlorobenzene at 30 °C. The CU analogues with no
phenolic hydroxyl groups did not retard oxidation of
either substrate, but CU and the three methoxy alky-
lphenols did inhibit oxidation, trapping four and two
peroxyl radicals, respectively. Moreover, the rate constant
(8) (a) Howard, J . A.; Ingold, K. U. Can. J . Chem. 1962, 40, 1851-
1864. (b) Burton, G. W.; Doba, T.; Gabe, E. J .; Hughes, L.; Lee, F. L.;
Prasad, L.; Ingold, K. U. J . Am. Chem. Soc. 1985, 107, 7053-7065.
(9) Valgimigli, L.; Banks, J . T.; Ingold, K. U.; Lusztyk, J . J . Am.
Chem. Soc. 1995, 117, 9966-9971.
(10) de Heer, M. I.; Mulder, P.; Korth, H.-G.; Ingold, K. U.; Lusztyk,
J . J . Am. Chem. Soc. 2000, 122, 2355-2360.
(11) Snelgrove, D. W.; Lusztyk, J .; Banks, J . T.; Mulder, P.; Ingold,
K. U. J . Am. Chem. Soc. 2001, 123, 469-477.
(12) Foti, M. C.; Barclay, L. R. C.; Ingold, K. U. J . Am. Chem. Soc.
2002, 124, 12881-12888.
(13) J ovanovic, S. V.; Steenken, S.; Boone, C. W.; Simic, M. G. J .
Am. Chem. Soc. 1999, 121, 9677-9681.
(19) Chignell, C. F.; Bilski, P.; Reszka, K. J .; Motten, A. G.; Sik, R.
H.; Dahl, T. A. Photochem. Photobiol. 1994, 59, 295-302.
(20) Gorman, A. A.; Hamblett, I.; Srinivasan, V. S.; Wood, P. D.
Photochem. Photobiol. 1994, 59, 389-398.
(21) For 2,4-pentanedione (acetylacetone), the [keto-enol]/[diketo]
ratios are 42, 4.8, 2.9, 1.2, and 0.23 in cyclohexane, 1,4-dioxane,
methanol, acetonitrile, and water, respectively, and equilibration is
extremely slow, e.g., 15 h in methanol. See: Mills, S. G.; Beak, P. J .
(14) For the 3 year period (2000 to 2002-end), Chemical Abstracts
(SciFinder) lists 956 papers with the word curcumin. Among this
number, 248 papers contain the word pair curcumin and antioxidant.
(15) J ovanovic, S. V.; Boone, C. W.; Steenken, S.; Trinoga, M.;
Kaskey, R. B. J . Am. Chem. Soc. 2001, 123, 3064-3068.
(16) Tønnesen, H. H.; Arrieta, A. F.; Lerner, D. Pharmazie 1995,
50, 689-693.
(17) Sugiyama, Y.; Kawakishi, S.; Osawa, T. Biochem. Pharmacol.
1996, 52, 519-523.
(18) Masuda, T.; Hidaka, K.; Shinohara, A.; Maekawa, T.; Takeda,
Y.; Yamaguchi, H. J . Agric. Food. Chem. 1999, 47, 71-77.
Org. Chem. 1985, 50, 1216-1224.
•
(22) The fastest authentic C-H abstractions appear to be: CH3
+
1,4-cyclohexadiene (1.3 × 105 M-1 s-1
(1.8 × 108 M-1 s-1).24
)
and Me3CO• + (CH3CH2)3N
23
J . Org. Chem, Vol. 69, No. 18, 2004 5889

Litwinienko and Ingold
SCHEME 1
for reaction of CU with polyperoxystyreneperoxyl radicals
(3.4 × 105 M-1 s-1) was exactly twice that of dehydroz-
ingerone (1.7 × 105 M-1 s-1), which is sometimes called
“half curcumin” (see DHZ in Chart 1), and the rate
constant for 2-methoxy-4-methylphenol was only slightly
lower (1.4 × 105 M-1 s-1). These results all demonstrate
that the two “halves” of CU react independently with
peroxyl radicals and that the R,γ-diketo moiety has no
antioxidant properties in these systems.
Barclay et al.’s6 paper and its conclusion “that cur-
cumin is a phenolic chain breaking antioxidant” were
both ignored a year later by J ovanovic et al.15 However,
in this second paper the original mechanism was modi-
fied with the proposal that “the initially generated
curcumin alkoxyl radical (see reaction 5) undergoes rapid
intramolecular H-shift” to form a phenoxyl radical; i.e.,
reaction 6.
molecule is unreactive toward radicals (for steric reasons).
The observed rate constants, ks, involve only noninter-
molecularly H-bonded phenol molecules. We demon-
strated that for any attacking radical and any HBD
substrate the value of ks in all solvents could be correlated
with the rate constant in a non-HB solvent, k°, via eq
711
log(ks/M-1 s-1) ) log(k°/M-1 s-1) - 8.3R2Hâ2H (7)
with some interesting exceptions.29 In this equation, RH2
represents the relative ability of the substrate to donate
a HB (range 0 to ca. 1)30 and âH2 represents the relative
ability of the solvent to accept a HB (range 0 to 1.00).31
The interesting exceptions referred to above were
discovered from the failure of eq 7 to predict HAT rate
constants for the reactions of various phenols with d p p h •
in alcohol solvents, the measured rates being higher
(sometimes very much higher) than predicted.29 To
explain these anomalies, we proposed a mechanism
which we now name as sequential proton loss electron
transfer (SPLET). In solvents which support ionization
(notably methanol among organic solvents29), the experi-
mental rate constant is the sum of the rate constant for
the conventional HAT process (Scheme 1, black) and the
very much larger rate constant for reaction of the radical
with the phenoxide anion,32,33 the SPLET process (Scheme
1, red). SPLET is favored for reactions of phenols having
low pKa’s with electron-deficient radicals having rela-
tively low HAT activities and yielding product molecules
having low pKas, e.g., d p p h •/d p p h -H and peroxyls, ROO•/
ROOH.34
This proposal was based on the fact that the radical
derived from a CU analogue with no phenolic hydroxyl
groups reacted with dioxygen (reaction 4), whereas the
CU-derived radical (like other phenoxyl radicals) did not.
Nevertheless, reaction 6 is implausible because an in-
tramolecular 1,9-H-atom shift could not possibly occur
in this planar radical.
Attempts to determine the most probable site of radical
attack on CU by theoretical calculations of bond dissocia-
tion enthalpies7b,25 have ignored a basic kinetic fact26 and,
hence, have not helped to resolve the controversy. To us,
it appeared reasonable to hypothesize that the disagree-
ments over CU’s reactive site arose from the different
experimental conditions that had been employed. In all
publications dealing with radical attack on CU, a direct
hydrogen atom transfer (HAT) has been assumed. Over
the past few years, we have shown that HAT rates from
phenols to any radical are subject to large kinetic solvent
effects (KSEs) in hydrogen-bond-accepting (HBA) sol-
vents.9-12,28 This is because a phenol molecule that is
involved as a hydrogen bond donor (HBD) to a solvent
The occurrence of SPLET in methanol and ethanol has
also been clearly demonstrated by Foti et al.32 in a study
of the reactions of d p p h • with some phenolic acids
(caffeic, p-coumaric, ferulic, and sinapic acids). Rate
constants for reactions of the methyl esters of these acids
(28) (a) Avila, D. V.; Ingold, K. U.; Lusztyk, J .; Green, W. H.;
Procopio, D. R. J . Am. Chem. Soc. 1995, 117, 2929-2930. (b) MacFaul,
P. A.; Ingold, K. U.; Lusztyk, J . J . Org. Chem. 1996, 61, 1316-1321.
(c) Valgimigli, L.; Ingold, K. U.; Lusztyk, J . J . Org. Chem. 1996, 61,
7947-7950. (d) Valgimigli, L.; Banks, J . T.; Lusztyk, J .; Ingold, K. U.
J . Org. Chem. 1999, 64, 3381-3383.
(23) Hawari, J . A.; Engel, P. S.; Griller, D. Int. J . Chem. Kinet. 1985,
17, 1215-1219.
(29) Litwinienko, G.; Ingold, K. U. J . Org. Chem. 2003, 68, 3433-
3438.
(24) Griller, D.; Howard, J . A.; Marriott, P. R.; Scaiano, J . C. J . Am.
Chem. Soc. 1981. 103, 619-623.
(30) Abraham, M. H.; Grellier, P. L.; Prior, D. V.; Duce, P. P.; Morris,
J . J .; Taylor, P. J . J . Chem. Soc., Perkin Trans. 2 1989, 699-711.
(31) Abraham, M. H.; Grellier, P. L.; Prior, D. V.; Morris, J . J .; Taylor
P. J . J . Chem. Soc., Perkin Trans. 2 1990, 521-529.
(32) Foti, M. C.; Daquino, C.; Geraci, C. J . Org. Chem. 2004, 69,
2309-2314.
(33) A very fast electron transfer from the phenolate anion generated
from 2,2,5,7,8-pentamethylchroman-6-ol to 2,2-bis-(4-tert-octylphenyl)-
1-picrylhydrazyl radical (DOPPH•) in acetonitrile has been recently
reported, see: Nakanishi, M.; Miyazaki, K.; Shimada, T.; Iizuka, Y.;
Inami, K.; Mochizuki, M.; Urano, S.; Okuda, H.; Ozawa, T.; Fukuzumi,
S.; Ikota, N.; Fukuhara, K. Org. Biomol. Chem. 2003, 1, 4085-4088.
(34) The pKa’s (in parentheses) for some relevant ion/molecule pairs
are:29 d p p h -, H+/d p p h H (8.5); ROO-, H+/ROOH (12.8); Me3CO-, H+/
Me3COH (19.2); primary alkyl radical, H+/alkane (ca. 50).
(25) (a) Wright, J . S. THEOCHEM 2002, 591, 207-217. (b) Sun,
Y.-M.; Zhang, H.-Y.; Chen, D.-Z.; Liu, C.-B. Org. Lett. 2002, 4, 2909-
2911.
(26) Thermochemistry alone does not dictate the relative rates of
H-atom abstractions from OH and CH groups. It is well-known that
for equal thermodynamic driving forces H-atom abstractions by oxygen-
centered radicals (and other radicals) are several orders of magnitude
faster from an OH group than from a CH group.27
(27) (a) Zavitsas, A. A. J . Am. Chem. Soc. 1972, 94, 2779-2789. (b)
Zavitsas, A. A.; Melikian, A. A. J . Am. Chem. Soc. 1975, 97, 2757-
2763. (c) Zavitsas, A. A.; Chatgilialoglu, C. J . Am. Chem. Soc. 1995,
117, 10645-10654, and the many references cited in these three
papers.
5890 J . Org. Chem., Vol. 69, No. 18, 2004

Abnormal Solvent Effects on Hydrogen Atom Abstraction
TABLE 1. Room -Tem p er a tu r e Bim olecu la r Ra te
Con sta n ts, ks (M-1 s-1) for th e Rea ction s of d p p h • w ith
Cu r cu m in (CU), Deh yd r ozin ger on e (DHZ), a n d
Isoeu gen ol (IE) in Va r iou s Solven ts a n d Acid ified
Solven tsa
were 3-5 times greater than for the free acids as a
consequence of the suppression of ionization of the
phenolic OH group by the free carboxylic acid. These
experiments nicely confirm the role that phenol ionization
can play in the reactions of phenols with d p p h • in
solvents that can support ionization.
CH3CO2Hc
solvent (âH2 , ꢀ)b
(mM)
ks (CU) ks (DHZ) ks (IE)
With this background information in our possession,
we hypothesized that the experimental conditions used
in some studies of CU’s antioxidant activity would favor
a purely HAT process (e.g., ref 6) while the conditions in
other studies would favor a large contribution from a
SPLET process (e.g., refs 13 and 15). Herein, we provide
evidence supporting this hypothesis and thus resolve the
CU antioxidant controversy. We also provide evidence
suggesting that CU’s R,γ-keto-enol/diketo moiety prob-
ably plays an important role in the reaction of CU with
d p p h • in solvents which support ionization.
1,4-dioxane (0.41,d 2.21)
dioxane
methanol (0.41, 32.63)
methanol
methanol
methanol
methanol
methanol
ethanol (0.45, 24.30)
ethanol
ethanol
ethanol
ethanol
ethanol
ethyl acetate (0.45, 6.02)
ethyl acetate
ethyl acetate
0
10
0
5
10
1.4
1.4
0.33
0.31
2.4
2.3
1000
12
7.4
5.5
5.4
6.4
240
10
7.2
16 000e 1450e
72
47
18
11
10
6.1
2.0
50
100
1000
0
5
10
1.0
3.6
0.72
9800e
59
790e
9.4
7.3
2.6
1.8
48
9.2
50
6.1
6.3
6.0
4.1
4.3
4.5
100
1000
0
10
1000
6.3
2.8
9.0
6.7
1.9
1.2
Resu lts
0.72
0.66
0.65
The d p p h • radical has served our mechanistic studies
of radical/phenol reactions well in the past9,11,12,28c,29 and
has been widely used to measure the hydrogen-atom-
donating abilities of natural antioxidants. The rates of
reaction 8 with XH () CU, IE, DHZ, 2-MeO-phenol and
2-MeO-4-Me-phenol) were determined by monitoring the
decay of d p p h • at 517 nm35 in a stopped-flow apparatus,
as described previously.29
a
ks values are generally based on at least two independent
measurements. Full data, including error limits for ks, are given
H
b
in the Supporting Information. â2 from ref 31, ꢀ from CRC
Handbook of Chemistry and Physics, 67th ed.; Weast, R. C., Ed.;
CRC Press: Boca Raton, FL, 1987. c â2H for propionic and butyric
acids ) 0.42, not available for acetic acid, and ꢀ for CH3CO2H )
d
6.15. This statistically corrected literature value is significantly
smaller than the value of 0.47 measured in the present work; see
text. e The loss of d p p h • follows excellent pseudo-first-order kinet-
ics in all cases except for CU and DHZ in neat MeOH and EtOH.
Thus, these four rate constants are based on initial rates of decay
of d p p h •.
d p p h • + XH f d p p h H + X•
(8)
The d p p h • concentration was generally ca. 2-6 × 10-5
M, and XH was used in excess. In almost all cases, this
produced excellent pseudo-first-order decays of d p p h •
(rate constant, kexptl) and the second-order rate constants
for reaction 8, ks, were calculated from the slopes of plots
of kexptl vs [XH], i.e.
acidified by the addition of (generally relatively low)
concentrations of acetic acid as in our earlier work in
which the SPLET reaction mechanism was first identi-
fied.29
In methanol, the d p p h •/IE reaction follows excellent
pseudo-first-order kinetics (Figure 1C) but the d p p h •/
CU and d p p h •/DHZ reactions exhibit a very fast initial
loss of d p p h • followed by a much slower process (Figure
1A,B). Similar results were obtained in ethanol. For all
three compounds in both alcohols the addition of as little
as 5 mM acetic acid dramatically decreases the reaction
rate and all the d p p h • decay traces now follow pseudo-
first-order kinetics (for acidified methanol solvent see
Figure 1A′-C′). With higher concentrations of acetic acid
the second-order rate constants in both alcohols decline
further toward a limiting value which is reached with
IE at [CH3CO2H] g 50 mM, with DHZ at [CH3CO2H] g
100 mM, and with CU at [CH3CO2H] g 1000 mM (see
Table 1). This behavior is fully consistent with a SPLET
reaction for these three substrates in both alcohols. That
is, these substrates are partially ionized in alcoholic
solvents and their anions react rapidly with the d p p h •
radical. Ionization is suppressed by addition of the acid,
and at sufficiently high [CH3CO2H], the SPLET process
is completely suppressed and the residual slow reaction
occurs by HAT.29 The three substrates are not expected
to ionize in dioxane (ꢀ ) 2.2), and therefore, as expected,
their second-order rate constants in this solvent are not
changed by the addition of acetic acid (Table 1).
kexptl ) ko + ks[XH]
(9)
at XH concentration where XH does not self-associate.
Equation 7 applies only to “pure” HAT reactions. It
therefore provides a vital mechanistic tool for quantifying
the importance of the SPLET process in different solvents
and with different phenols.29 Unfortunately, k° in eq 7
could not be determined for the d p p h •/CU reaction
because of CU’s low solubility in saturated hydrocarbons
(âH2 ) 0.00). We therefore measured ks for this reaction
in two non-hydroxylic solvents, 1,4-dioxane and ethyl
acetate, and in two hydroxylic solvents, methanol and
ethanol. Dioxane and methanol have been reported to
have identical HBA activities (i.e., identical âH2 values )
0.41).31 Similarly, ethyl acetate and ethanol have also
been reported to have identical HBA activities (â2H
)
0.45).31 However, the two hydroxylic solvents have much
higher dielectric constants, ꢀ, than the two non-hydroxylic
solvents and hence have much greater abilities to support
ionization and, consequently, the SPLET reaction mech-
anism (Scheme 1). Results are summarized in Table 1
together with ks values measured in these four solvents
(35) CU is yellow with a strong absorption in the 420 to 430 nm
region in organic solvents (ꢀλmax ∼ 55 000).16 In none of our work did
CU’s absorption interfere with our measurements of the rates of decay
of d p p h • at 517 nm (ꢀλmax ∼ 50 000).
In ethyl acetate, with its higher dielectric constant
(ꢀ ) 6.0), the addition of CH3CO2H does not change the
J . Org. Chem, Vol. 69, No. 18, 2004 5891

Litwinienko and Ingold
F IGURE 1. Decay of 3 µM d p p h • in reaction with 0.37 mM CU, DHZ, and IE in MeOH, panels A, B, and C, respectively, and
in MeOH containing 5 mM acetic acid, panels A′, B′, and C′, respectively. A ) absorbance, Ao ) initial absorbance.
TABLE 2. Va lu es of p Ka in Wa ter /Meth a n ol (1:1 v/v)
second-order rate constants for DHZ and IE but it does
fr om th e Liter a tu r e a n d Mea su r ed in th e P r esen t Wor k a
reduce the rate constant for CU (Table 1). Clearly, CU
must be a stronger acid than DHZ and IE and be
pKa
partially ionized in ethyl acetate. This is consistent with
pKa values in 1:1 (v/v) water/methanol obtained from the
literature for CU36,37 and dimethoxycurcumin and deter-
mined in the present work for DHZ, IE, and two related
phenols, 2-MeO-phenol and 2-MeO-4-Me-phenol (see
Table 2). More complete pKa data are given in the
Supporting Information.
The limiting HAT rate constants (i.e., k’s when SPLET
has been suppressed, if necessary, by acetic acid) for the
reactions of d p p h • with CU, DHZ, and IE are all smaller
in dioxane than in ethyl acetate, methanol, and ethanol
CU
(1)
(2)
(3)
8.54b
9.30b
10.69b
8.75b
9.12
10.60
10.68
10.78
(Me2CU)c
DHZ
IE
2-MeOC6H4OH
2-MeO-4-MeC6H3OH
a
b
For additional details, see the Supporting Information. See
refs 36 and 37. c Dimethoxycurcumin is the dimethyl ether of CU
having no phenolic OH groups.
(see Table 1). Such kinetics appeared to be inconsistent
with the literature â2H values31 for these solvents (Table
1) which imply (equation 7) that HAT rate constants
should, for example, be smaller in ethyl acetate than in
dioxane. This caused us to redetermine (unnecessarily,
see below) âH2 for dioxane by the usual infrared (IR)
spectroscopic method31,38 using CCl4 as solvent and
“calibrated” 4-fluorophenol as the HBD (for details see
(36) Borsari, M.; Ferrari, E.; Grandi, R.; Saladini, M. Inorg. Chim.
Acta 2002, 328, 61-68.
(37) All three pKa’s for CU are within 2.5 log units of one another
and the three ionization processes therefore overlap. We were not able
to deconvolute the CU potentiometric curve. Fortunately, Bosari et
al.36 had reported some very systematic studies of CU’s acidity using
potentiometric, spectrophotometric, and NMR methods. The three pKa
of CU were assigned by these workers to the ionization of particular
OH groups, see Discussion.
5892 J . Org. Chem., Vol. 69, No. 18, 2004

Abnormal Solvent Effects on Hydrogen Atom Abstraction
TABLE 3. Com p a r ison of Mea su r ed Bim olecu la r Ra te Con sta n ts (Bold ) for th e Rea ction s of d p p h • a n d F ou r
o-Meth oxyp h en ols in Eth yl Aceta te a n d Dioxa n e w ith Ra te Con sta n ts Ca lcu la ted (Ita lics) fr om Eq 7a
k° (M-1 s-1) (in heptane)
âH2 ) 0.00b
ks (M-1 s-1) (in ethyl acetate)
âH2 ) 0.45b
ks (M-1 s-1) (in dioxane)
phenol RH2
âH2 ) 0.47c
(âH2 ) 0.41)b
c
DHZ
IE
2-MeO
2-MeO-4-Me
0.361
0.291
0.294
7.2d
53
1.0
0.72
4.1
0.14
0.89
0.33
4.4
0.08
0.59
0.33
2.4
0.055
0.47
0.28
3.9
0.07
0.52
(0.43)
(5.5)
(0.10)
(0.73)
e
0.292
7.1
a
b
d
Full data with errors are given in the Supporting Information. Reference 31. c This work. In heptane/CCl4, (1:1 v/v) because of the
low solubility of DHZ in neat heptane. e A value of 0.26 is given in ref 30.
the Experimental Section and Supporting Information).
The âH2 value obtained for dioxane was 0.47, which is
fully consistent with the HAT reactions of CU, DHZ, and
IE being slightly slower in dioxane than in ethyl acetate
(âH2 ) 0.45).31 Later, a reviewer pointed out that in
Abraham et al.’s original report31 the log KBH value for
dioxane is listed as 1.101. When this was converted to
a âH2 value, a statistical factor of 2 was introduced,
giving âH2 ) 0.41.31 If the statistical factor is ignored, as
we believe it should be, âH2 for dioxane becomes 0.475, in
excellent agreement with our own measurements.
The validity of eq 7 for intramolecular hydrogen-
bonded compounds was also briefly explored using DHZ,
IE, 2-methoxyphenol, and 2-methoxy-4-methylphenol.
Unfortunately, CU had to be excluded from these studies
because it was found to be too insoluble in CCl4 and
heptane. First, the RH2 values for these four ortho-
methoxyphenols were measured by the usual IR
method,11,12,29,30 using CCl4 as solvent and the “calibrated”
HBA, DMSO (for details, see the Supporting Infor-
mation). The RH2 values for IE, 2-methoxyphenol, and
2-methoxy-4-methylphenol are all 0.29 (see Table 3) but
that for DHZ is appreciably greater (0.36). These results
are consistent with the observation that DHZ is signifi-
cantly more acidic than the other three o-methoxyphenols
(see Table 2). The greater acidity of DHZ and its stronger
HBD activity can be attributed to the electron-withdraw-
ing effect of its carbonyl group which is conjugated to the
aromatic ring para to the hydroxyl group (see Chart 1).
Second, the rate constants for the reactions of these four
o-methoxyphenols with d p p h • were measured in heptane
and for 2-MeO- and 2-MeO-4-Me-phenols in ethyl acetate
and dioxane; see Table 3. Values of ks in these two
solvents were then calculated using eq 7 and the k°, RH2 ,
and âH2 values given in Table 3. These calculated values
are in very satisfactory agreement with our measure-
ments (see Table 3). Notably, the calculated ks values in
dioxane agree better with experiment using the â2H
value (0.47) measured in the present work instead of the
statistically corrected literature value (0.41).
low pKa (as is the case for d p p h •/d p p h -H (8.5) and ROO•/
ROOH (12.8).29 (ii) The substrate, ArOH, must have a
readily “abstractable” and fairly acidic hydrogen atom (as
is true for most phenols: ArOH h ArO- + H+; ArOH +
X• h ArO• + XH). (iii) The solvent must be able to support
partial (or even complete) ionization of the substrate.
The simplest test we have devised which can demon-
strate the presence or absence of a SPLET contribution
to an (apparent) HAT is to study the effect of added acetic
acid on the measured rate constant for the reaction. The
ionization of most substrates will be suppressed by added
acetic acid (because the substrates will generally be less
acidic than CH3CO2H). As a consequence, the contribu-
tion of SPLET to the overall reaction will be decreased
and the measured rate constant will decline. In many
cases, SPLET will be completely suppressed at high [CH3-
CO2H] (see ref 29). The measured rate constant will then
become independent of the concentration of CH3CO2H
and will reflect the underlying and much slower HAT
process in the solvent in which the reaction was carried
out. To prove that these limiting rate constants really
do correspond (at least, mainly) to the underlying HAT
reaction in any particular solvent it is necessary to show
that they are approximately of the magnitude dictated
by eq 7.29,39 In the present work, this was achieved by
demonstrating that the “limiting” rate constants at high
[CH3CO2H] in the two ionizing solvents, MeOH and
EtOH, were roughly equal to rate constants for the same
ArOH/X• reaction measured in solvents having similar
HBA activities (i.e., similar âH2 values) and little or no
ability to ionize ArOH. To confirm that these low dielec-
tric constant solvents (dioxane and ethyl acetate in the
present work) truly prevented SPLET it was, of course,
essential to also measure rate constants with acetic acid
added to these solvents. This check confirmed that CU,
DHZ, and IE were not ionized in dioxane (Table 1) and
therefore SPLET played no role in this solvent. The same
is true for IE and, probably, for DHZ in ethyl acetate,
but CU must be partially ionized because the addition
of 1 M CH3CO2H reduced the rate constant for reaction
with d p p h • by a factor of 4.6 from its value in neat ethyl
acetate (Table 1).
First, consideration will be given to HAT from a
phenolic hydroxyl group of CU, DHZ, and IE to d p p h •.
These reactions are dominant in dioxane, in strongly
acidified methanol and ethanol, and in ethyl acetate, the
last strongly acidified for CU. A comparison of the rate
constants measured under these conditions shows that
Discu ssion
For an apparent hydrogen atom abstraction (HAT)
reaction between X• and ArOH (reaction 1) to occur by a
sequential proton loss electron transfer (SPLET) mech-
anism, three conditions must be met: (i) The radical, X•,
must be electron deficient; i.e., XH must have a fairly
(39) Equation 7 is a remarkably reliable kinetic guideline for “pure”
HAT processes (see e.g., Table 3) but it appears to be rarely, if ever,
precisely obeyed (see e.g., refs 11, 12, and 29). Too much should not,
therefore, be read into any minor kinetic “inconsistencies”.
(38) (a) Besseau, F.; Laurence, C.; Berthelot, M. J . Chem. Soc.,
Perkin Trans. 2 1994, 485-489. (b) Besseau, F.; Luc¸on, M.; Laurence,
C.; Berthelot, M. J . Chem. Soc., Perkin Trans. 2 1998, 101-107.
J . Org. Chem, Vol. 69, No. 18, 2004 5893

Litwinienko and Ingold
IE is the most reactive HAT substrate, followed by CU
(even after statistically correcting for the two phenolic
groups in this molecule), followed by DHZ. This order of
reactivity is congruent with expectations based on the
fact that electron withdrawing (EW) para substituents
increase phenolic O-H bond dissociation enthalpies
(BDEs)40 and hence reduce HAT activities.11,41 The EW
carbonyl group in DHZ is conjugated to the aromatic
ring. If CU existed solely in the diketo form, 1a , it would
(after the statistical correction) be of equal reactivity to
DHZ. However, in these four solvents/solvent mixtures,
CU probably exists largely as the keto-enol,19-21 1b, and
the -(O)CCHdC(OH)- moiety is expected to be less
strongly EW than the simple carbonyl group in DHZ. For
this reason, even after the statistical correction, CU is a
somewhat better HAT agent to d p p h • than DHZ. The
absence of a carbonyl group in the para substituent of
IE makes this compound a more reactive HAT than CU
and DHZ. IE is also a more reactive HAT agent than
2-methoxy-4-methylphenol which, in turn, is more reac-
tive that 2-methoxyphenol (Table 3). These results are
also congruent with expectations. That is, para substit-
uents which aid electron delocalization in phenoxyl
radicals decrease phenolic O-H BDEs and hence increase
HAT rates. For these three phenols, the order of increas-
ing HAT rates is, as expected, 4-H < 4-CH3 < 4-CHd
CHCH3. DHZ is more interesting. The HAT activity of
DHZ is essentially identical to that of 2-methoxy-4-
methylphenol. This implies that the O-H bond weaken-
ing effect due to electron delocalization into the 4-CHd
CHCOCH3 group is countered by bond strengthening
arising from the EW nature of this group.42 The EW
character of 4-CHdCHCOCH3 is attested to by the much
lower pKa of DHZ than IE, 2-methoxy- and 2-methoxy-
4-methylphenol and by a reported σ+p (CHdCHCOCH3)
) 0.39.43
must be faster than its reaction with ca. 3 µM d p p h •. In
contrast, the reactions of CU and DHZ in the neat
alcohols show an initial, very fast reaction followed by a
slow (but still rapid) loss of d p p h • (Figure 1A,B and
footnote e in Table 1). These initial fast reactions must
correspond to rapid reactions of the d p p h • with the CU
and DHZ anions present at low equilibrium concentra-
tion in the alcohols. These “preformed” anions are quickly
depleted and ionization of the much larger concentration
of neutral CU and DHZ becomes partly rate limiting.44
The occurrence of SPLET in the reactions of CU, DHZ,
and IE with d p p h • is perfectly reasonable as judged by
our earlier work29 because all three substrates are fairly
acidic with CU being the strongest acid. The three pKa’s
of CU have been determined in water/methanol, 1:1 (v/
v) as36 (i) 8.54 for dissociation of the R,γ-keto-enol, (ii)
9.30 for dissociation of the first phenolic hydroxyl group
to give the dianion, and (iii) 10.69 for dissociation of the
second phenolic hydroxyl and formation of the trianion
(see Table 2 and, for more details, the Supporting
Information).
The low first pKa value for the keto-enol moiety in CU
indicates that in solvents which support ionization the
SPLET mechanism most probably involves proton loss
from (i.e., ionization of) the R,γ-keto-enol moiety and
electron transfer from the resulting [(O)CCHC(O)]- anion
moiety to the d p p h • radical. That such a reaction can
occur was demonstrated using 2,4-pentadione (acetylac-
etone, AcAc). In dioxane, the reaction of d p p h • with AcAc
followed good pseudo-first-order kinetics but was very
slow (k ) (8 ( 4) × 10-5 M-1 s-1) and is presumably a
HAT reaction. In methanol the initial reaction was much
faster (k ) (4 ( 1) × 10-2 M-1 s-1) but subsequently the
reaction became slower and the overall reaction did not
follow pseudo-first-order kinetics. Thus, in this alcohol,
the initial fast reaction probably involves electron trans-
fer from the AcAc anion to the d p p h •.
The extremely high rate constants for the reactions of
CU, DHZ, and IE with d p p h • in methanol and ethanol
compared with the rate constants in dioxane and ethyl
acetate, taken together with the dramatic declines in the
rate constants in the two alcohols which are produced
by added acetic acid, prove that in the alcohols these
reactions occur primarily by the SPLET mechanism. As
was pointed out in the Results, the reaction of IE in the
neat alcohols follows clean pseudo-first-order kinetics (see
Figure 1C and Table 1). This means the ionization of IE
In the case of CU, the initial electron transfer from
the ionized keto-enol moiety (see structure II in Scheme
2) to the d p p h • produces a radical fragment [(O)CCHC-
(O)]• (structure III), which will be strongly electron
withdrawing.46 Consequently, the pKa of the phenolic
hydroxyl groups will decrease dramatically, and this will
lead to proton loss from one of these groups (structure
(40) (a) Mulder, P.; Saastad, O. W.; Griller, D. J . Am. Chem. Soc.
1988, 110, 4090-4092. (b) J onsson, M.; Lind, J .; Eriksen, T. E.;
Merenyi, G. J . Chem. Soc., Perkin Trans. 2 1993, 1567-1568. (c)
Wayner, D. D. M.; Lusztyk, E.; Ingold, K. U.; Mulder, P. J . Org. Chem.
1996, 61, 6430-6433. (d) Dorrestijn, E.; Laarhoven, L. J . J .; Arends,
I. W. C. E.; Mulder, P. J . Anal. Appl. Pyrol. 2000, 54, 153-192. (e)
Pratt, D. A.; de Heer, M. I.; Mulder, P.; Ingold, K. U. J . Am. Chem.
Soc. 2001, 123, 5518-5526. (f) Pratt, D. A.; DiLabio, G. A.; Mulder,
P.; Ingold, K. U. Acc. Chem. Res. 2004, 37, 334-340.
(44) A truly rate-limiting ionization would cause the slow second
stage of d p p h • loss to follow zero-order kinetics in [d p p h •] while
remaining first order in substrate. This phenomenon has been observed
in other d p p h •/ArOH reactions.45 Rates of d p p h • loss in the slow
second stages of its reaction with CU and DHZ in the neat alcohols
indicate that the kinetics of ionization and of the d p p h •/anion reactions
are convoluted. Convolution was least for DHZ in methanol and in
this case the equilibrium concentration of the anion, [DHZ-]eq, was
estimated as follows. At constant [DHZ] (0.98 mM) and variable
[d p p h •] (8-91 µM) the roughly linear slow second stages of d p p h •
decay were extrapolated back to zero time to obtain [d p p h •]intercept. The
differences between the initial concentrations of d p p h •, [d p p h •]o, and
[d p p h •]intercept was plotted against 1/[d p p h •]o. A reasonable straight
(41) (a) Howard, J . A.; Ingold, K. U. Can. J . Chem. 1963, 41, 1744-
1751. (b) Howard, J . A.; Ingold, K. U. Can. J . Chem. 1963, 41, 2800-
2806.
(42) The near equivalence of these two phenolic O-H BDEs is
supported by DFT calculations of the differences in O-H BDEs: (4-
CH3COCHdCHC6H4O-H
- C6H5O-H) and (4-CH3C6H4O-H -
C6H5O-H) which are -1.8 and -2.6 kcal/mol, respectively. These
calculations were performed using the B3LYP functional with the
medium-level model MLM1 as described in: DiLabio, G. A.; Pratt, D.
A.; LoFaro, A. D.; Wright, J . S. J . Phys. Chem. A 1999, 103, 1653-
1661.
line was obtained with an intercept at 1/[d p p h •]o ) 0 of 2.6 × 10-6
M
(see Supporting Information). This should correspond to [DHZ-]eq and
the percentage DHZ ionized is given by (100 × 2.6 × 10-6)/[(980-2.6)
× 10-6] ) 0.27%.
(43) Saldabol, N. O.; Popelis, Yu. Yu.; Liepin’sh, E. E. Zh. Org. Khim.
1980, 16, 1494-1497 (English translation, pp 1285-1288).
(45) Litwinienko, G. Unpublished results.
5894 J . Org. Chem., Vol. 69, No. 18, 2004

Abnormal Solvent Effects on Hydrogen Atom Abstraction
SCHEME 2
nitrogen purging when necessary, until they were taken up
into the glass syringes of the stopped-flow apparatus with their
gastight Teflon plungers. The decay of d p p h • in the presence
of known concentrations of the phenols was followed at 517
nm on an Applied Photophysics stopped-flow spectrophotom-
eter, SX 18 MV, equipped with a 150 W xenon lamp. All
measurements were carried out at 23 ( 2 °C. The concentra-
tion of d p p h • was (4 ( 2) × 10-5 M, and the phenols were
always used in large excess. The decays of the d p p h • absor-
bancies were analyzed as pseudo-first-order processes to yield
IV which will evolve to V by migration of the negative
charge, see Scheme 2).
That is, in solvents that support ionization, CU reacts
with electrophilic radicals initially at the ionized R,γ-
keto-enol moiety and the resulting neutral radical loses
a phenolic proton, thus yielding the same phenoxyl
radical, V, as would have been formed by HAT from the
phenolic hydroxyl group of the CU anion (II) to the
radical (Scheme 2). It must be emphasized that phenoxyl
radical formation in ionizing solvents does not occur by
J ovanovic et al.’s15 “rapid intramolecular H-shift”, reac-
tion 6, but by loss of a proton into the bulk solution from
the neutral [(O)CCHC(O)]• CU radical, III. Of course, in
solvents which do not support ionization or in which
ionization is prevented by the addition of CH3CO2H (or
other proton source) the SPLET mechanism cannot occur
and the reactions will involve only HAT from a phenolic
hydroxyl group of the neutral CU to the radical. This has
already been clearly demonstrated by Barclay et al.6 in
the solvent, chlorobenzene. Thus, the CU antioxidant
controversy is resolved.
k
exptl/s-1. For the concentrations of phenols used in our experi-
ments (see Tables S1-S23) the plots of kexptl vs phenol
concentration were linear and their slopes gave the second-
order rate constants, ks. Mean values ks with absolute errors
(∆k) and statistical parameters are provided in the Supporting
Information.
Deter m in a tion of R2H a n d â2H. Values of RH2 for substrates
were determined by monitoring the OH fundamental stretch-
ing region of their IR spectrum in CCl4 at ambient tempera-
tures in the usual manner.11,29 Each spectrum was collected
on a Shimadzu FTIR 8201PC apparatus using a 1.03 mm CaF2
cell (20 scans with a resolution of 1 cm-1), and the baseline
was corrected using solutions having the same concentrations
of hydrogen bond acceptor (HBA). Although the OH groups in
o-methoxyphenols are internally hydrogen bonded, the forma-
tion of intermolecular HB complex with HBA is still described
by
Exp er im en ta l Section
Ma ter ia ls. 1,4-Dioxane (99+%) and ethyl acetate (99.99%)
were used as received. Values of ks in methanol and ethanol
were extremely sensitive to traces of base, the amounts of
which vary even in samples of the highest commercially
available purity (according to the labels on the bottles). These
alcohols were, therefore, fractionally distilled from a small
amount of d p p h • (ca. 1 mg per 500 mL of alcohol) and a few
beads of a weakly acidic ion-exchange resin. IE (98%, a
mixture of cis and trans isomers), CU (97%), 2-MeO-phenol
(99%), 4-Me-2-MeO-phenol (98%), and 4-fluorophenol (99%)
were commercial materials that were used without further
purification. Highly purified DHZ was a gift; see the Acknowl-
edgment.
ArOHfree + HBAfree h ArOH‚‚‚HBAintermolecular
and the equilibrium constant is given by
[ArOH‚‚‚HBA]intermolecular
Ki )
(11)
[ArOH]free[HBA]free
where [ArOH]free denotes the concentration of phenol molecules
not participating in an intermolecular HB. Values of [ArOH]free
were determined from the decrease in the peak height of the
internally hydrogen bonded OH at about 3560 cm-1 for 2-MeO-
phenol and 2-MeO-4-Me-phenol, 3557 cm-1 for IE and 3549
cm-1 for DHZ (see Figures S5 and S6 in the Supporting
Information). Equation 11 can be transformed to the form
[ArOH]o/[ArOH]free) 1 + Ki [HBA]free, where [ArOH]o is the
total concentration of phenol and values of Ki were determined
Kin etic Mea su r em en ts. The procedure used to determine
ks was generally the same as described previously.29 Briefly,
solutions of d p p h • and the phenols were prepared in nitrogen-
purged solvents and were kept under nitrogen, with additional
(46) The σp (or σ+p ) value for O• has been estimated to be ca. 2.047
though a lower value, closer to that of the strongly electron withdraw-
ing (EW) NO2 group (σp ) 0.78), appears to be more probable.48 In
either event the O• moiety is a powerful EW group and the EW
character of the [(O)CCHC(O)]• group will therefore be substantial even
though it is likely to be somewhat attenuated relative to O• by
delocalization of the unpaired electron over the atoms in this group
and by conjugation through a vinyl group.
from the plots of the ratio [ArOH]o/[ArOH]free versus [HBA]free
.
Values of RH2 were calculated using the equations
log KHA ) (log Ki - DB)/LB
(12)
i
and
J . Org. Chem, Vol. 69, No. 18, 2004 5895

Litwinienko and Ingold
RH2 ) (log KHA + 1.1)/4.636
(13)
i
mental data and calculations are given in the Supporting
Information.
The HBA was DMSO, a calibrated base for which LB ) 1.24
and DB ) 0.266.30
Ack n ow led gm en t. We are grateful to Dr. L. R. C.
Barclay for a generous gift of DHZ. We thank Erin R.
J ohnson, Dr. Gino A. DiLabio, and Dr. H.-G. Korth for
helpful discussions and two anonymous reviewers for
some extremely valuable comments, despite all the extra
work these comments required of us. G.L. thanks the
Natural Science and Engineering Research Council of
Canada (NSERC) for a one year Visiting Fellowship.
Values âH2 for dioxane were calculated from the equation
âH2 ) (log KHB + 1.1)/4.636
(14)
where log KHB is connected to the experimental equilibrium
constants Ki by the equation: log Ki ) LA log KH + DA (see ref
B
31). For the reference acid, 4-fluorophenol, LA ) 1.000 and DA
) 0.000,31 and hence log KHB ) log Ki. For dioxane plus
4-fluorophenol values of Ki ) 11.58 and 11.68 were obtained,
from which the value of âH2 is calculated to be 0.47.
Su p p or tin g In for m a tion Ava ila ble: Detailed kinetic
data for reactions of the CU, DHZ, IE, 2-MeO-phenol and
4-Me-2-MeO-phenol with d p p h • in neat heptane and in neat
and acidified dioxane, methanol, ethanol, and ethyl acetate
(Tables S1-S23) and for reaction of AcAc with d p p h • in neat
methanol and in dioxane (Tables S9, S14, S22, and S23), plots
of log ks vs [CH3CO2H] in methanol for the reaction of CU +
d p p h • (Figure S1), dependence of the experimental first-order
rate constant, kexptl, vs concentration of [AcAc] in methanol
(Figure S2), examples of d p p h • decay traces for fixed initial
concentration of DHZ and variable initial concentration of
d p p h • in methanol, plot of ∆[d p p h •] as a function of reciprocal
of [d p p h •]o, (Figures S3 and S4), IR spectra of IE and DHZ in
CCl4 containing various concentrations of DMSO (Figures S5
and S6), plots and parameters used for calculation of the HB
equilibrium constants and the RH2 values for the phenols and
âH2 for dioxane (Figures S7 and S8 and Tables S24-S26),
experimental parameters used for the determination of the
ionization constants of the phenols in water-methanol (1:1, v/v)
solution (Tables S27-S35), and experimental and literature
values of pKa (Table S36). This material is available free of
charge via the Internet at http://pubs.acs.org.
Deter m in a tion s of p Ka . Experimental pKa values for all
the phenols (except CU) in water-methanol (1:1) were deter-
mined by potentiometric titration: pKa ) pH + log([ArOH]/
[ArO-]) with correction for the OH- activity (if required)
according to the procedure described by Albert and Sergeant.49
A universal pH-meter, CX-731Elmetron, was used with a
combined pH glass electrode calibrated on primary pH stan-
dards for nonaqueous and mixed solvents as recommended by
IUPAC.50 A constant ionic strength (0.1 M KCl) was main-
tained in all experiments. The titrant (KOH in water-
methanol 1:1 mixture containing a small quantity of BaCl2 to
remove traces of carbonates) and titrated solutions were kept
in atmospheres free from carbon dioxide. Tables with experi-
(47) Pratt, D. A.; DiLabio, G. A.; Valgimigli, L.; Pedulli, G. F.; Ingold,
K. U. J . Am. Chem. Soc. 2002, 124, 11085-11092.
(48) Korth, H.-G. Private communication.
(49) Albert, A.; Serjeant, E. P.: The determination of ionization
constants - Laboratory Manual; Chapman and Hall: New York, 1984.
(50) IUPAC Compendium of Analytical Nomenclature, 3rd ed.;
Inczedy, J ., Lengyel, T., Ure, A. M., Eds.; Blackwell Science: Cam-
bridge, MA 1998 (online version: http://www.iupac.org/publications/
analytical_compendium/.
J O049254J
5896 J . Org. Chem., Vol. 69, No. 18, 2004