Hydroxylamines as oxidation catalysts: Thermochemical and kinetic studies

Article abstract of DOI:10.1021/jo026660z

Bond dissociation enthalpies (BDE) of hydroxylamines containing alkyl, aryl, vinyl, and carbonyl substituents at the nitrogen atom have been determined by using the EPR radical equilibration technique in order to study the effect of the substituents on th

Full text of DOI:10.1021/jo026660z

Hyd r oxyla m in es a s Oxid a tion Ca ta lysts: Th er m och em ica l a n d
Kin etic Stu d ies
Riccardo Amorati, Marco Lucarini, Veronica Mugnaini, and Gian Franco Pedulli*
Dipartimento di Chimica Organica “A. Mangini”, Universita` degli Studi di Bologna,
Via S. Donato 15, 40127 Bologna, Italy
Franceso Minisci and Francesco Recupero
Dipartimento di Chimica - Politecnico di Milano, Via Mancinelli 7, 20131 Milano, Italy
Francesca Fontana
Dipartimento di Ingegneria, Universita` di Bergamo, Viale Marconi 5, 24044 Dalmine (BG), Italy
Paola Astolfi and Lucedio Greci
Dipartimento di Scienza dei Materiali e della Terra, Universita` di Ancona, Via Brecce Bianche,
60131 Ancona, Italy
pedulli@alma.unibo.it
Received November 5, 2002
Bond dissociation enthalpies (BDE) of hydroxylamines containing alkyl, aryl, vinyl, and carbonyl
substituents at the nitrogen atom have been determined by using the EPR radical equilibration
technique in order to study the effect of the substituents on the O-H bond strength of these
compounds. It has been found that substitution of an alkyl group directly bonded to the nitrogen
atom with vinyl or aryl groups has a small effect, while substitution with acyl groups induces a
large increase of the O-H BDE value. Thus, dialkyl hydroxylamines have O-H bond strengths of
only ca. 70 kcal/mol, while acylhydroxylamines and N-hydroxyphthalimide (NHPI), containing two
acyl substituents at nitrogen, are characterized by BDE values of ca. 80 and 88 kcal/mol, respectively.
Since the phthalimide N-oxyl radical (PINO) has been recently proposed as an efficient oxidation
catalyst of hydrocarbons or other substrates, the large BDE value found for the parent hydroxyl-
amine (NHPI) justifies this proposal. Kinetic studies, carried out in order to better understand the
mechanism of the NHPI-catalyzed aerobic oxidation of cumene, are consistent with a simple kinetic
model where the rate-determining step is the hydrogen atom abstraction from the hydroxylamine
by cumylperoxyl radicals.
SCHEME 1
In tr od u ction
The oxidation and, in general, the functionalization of
organic substrates, such as aliphatic hydrocarbons, is a
primary essential tool in organic synthesis and in indus-
trial chemistry. Recently, it has been shown that N-
hydroxyphthalimide (NHPI) is an effective catalyst for
the oxidation of organic compounds by molecular oxygen
under mild conditions.1,2 This reaction is believed to occur
(see Scheme 1) via the intermediation of carbon-centered
radicals from the substrate produced by the phthalimide
(1) Yoshino, Y.; Hayashi, Y.; Iwahama, T.; Sakaguchi, S.; Ishii, Y.
J . Org. Chem. 1997, 62, 810. Iwahama, T.; Yoshima, Y.; Keitoku, T.;
Sagakuchi, S.; Ishii, Y. J . Org. Chem. 2000, 65, 6502. Sakaguchi, S.;
Nishiwaki, Y.; Kitamura, T.; Ishii, Y. Angew. Chem., Int. Ed. 2001,
40, 222. Ishii, Y.; Sakaguchi, S.; Iwahama, T. Adv. Synth. Catal. 2001,
343, 393.
(2) Cecchetto, A.; Fontana, F.; Minisci, F.; Recupero, F. Tetrahedron
Lett. 2001, 42, 6651. Minisci, F.; Punta, C.; Recupero, F.; Fontana, F.;
Pedulli, G. F. J . Org. Chem. 2002, 67, 2671. Cecchetto, A.; Minisci, F.;
Recupero, F.; Fontana, F.; Pedulli, G. F. Tetrahedron Lett. 2002, 43,
3605. Minisci, F.; Punta, C.; Recupero, F.; Fontana, F.; Pedulli, G. F.
Chem. Commun. 2002, 7, 688-689.
N-oxyl radical (PINO), generated in situ. Under aerobic
conditions, the resulting alkyl radicals are readily trapped
by molecular oxygen to give peroxyl radicals that in turn
abstract the hydrogen atom from NHPI regenerating
PINO and giving rise to a hydroperoxide.
10.1021/jo026660z CCC: $25.00 © 2003 American Chemical Society
Published on Web 02/01/2003
J . Org. Chem. 2003, 68, 1747-1754
1747

Amorati et al.
TABLE 1. EP R Sp ectr a l P a r a m eter s of Nitr oxid es 1b-8b a n d P INO a n d BDE Va lu es of th e O-H Bon d in th e P a r en t
Hyd r oxyla m in es 1a -7a a n d NHP I
radical
solventa
aN/gauss
other/gauss
g factor
BDE/kcal mol-1
1b
2b
3b
4b
5b
6b
7b
8b
B
B
B
B
B
9.28
10.80
10.88
5.72
5.82
7.20
7.25
7.43
4.36
3.09 (1H), 2.89 (1H) 1.03 (2H)
2.71 (2H), 1.03 (2H)
1.03 (2H)
1.57 (2H)
0.79 (1H), 0.67 (1H)
8.05 (3H)
2.0058
2.0058
2.0056
2.0055
2.0062
2.0065
2.0068
2.0065
2.0073
70.6 ( 0.3
71.4 ( 0.3
69.7 ( 0.4
72.8 ( 0.4
78.5 ( 0.5
79.2 ( 0.5
80.2 ( 0.5
-
B/BuOH (10:1)
B/BuOH (10:1.5)
B
1.53 (3H), 0.64 (2H)
0.45 (2H)
PINO
BuOH
88.1 ( 0.6
a
B ) benzene.
According to this mechanism, the key step of the
overall reaction is the abstraction by PINO of a hydrogen
atom from RH (Scheme 1, eq 1) that will proceed at an
acceptable rate only if the value of the bond dissociation
enthalpy (BDE) of the O-H bond formed is higher or
similar to the strength of the C-H bond cleaved, thus
making the knowledge of the BDE of NHPI of great
importance. Since the C-H BDE values in the great
majority of organic compounds are in the range 85-100
kcal/mol3 while the only experimentally determined O-H
BDE values measured of hydroxylamines are close to 70
kcal/mol,4 reaction 1 seems strongly endothermic and
thus hardly feasible on thermodynamic grounds. It
should be pointed out, however, that BDE measurements
have only been reported for dialkylhydroxylamines while
NHPI contains two carbonyl substituents adjacent to the
NOH group. Since these might induce a strengthening
of the O-H bond with respect to alkyl substituents, we
have undertaken a thermochemical study on NHPI and
other hydroxylamines in order to investigate the effect
of alkyl, aryl, vinyl, and carbonyl substituents on the
O-H bond strength. The investigated hydroxylamines,
beside NHPI, are those shown below.
hydrogen atom abstraction from NHPI by peroxyl radi-
cals, the effect of the addition of the hydroxylamine on
the rate of the thermally initiated autoxidation of cumene
was investigated.
Resu lts a n d Discu ssion
Th er m och em istr y. The determinations of the bond
dissociation enthalpies of hydroxylamines were done by
using the EPR radical equilibration technique that has
been largely used in our laboratory in order to measure
BDE values of the O-H bond in phenols5,6 and of the
N-H bond in aromatic amine antioxidants.7 This tech-
nique consists of measuring the equilibrium constant for
the hydrogen atom transfer reaction between a hydroxy-
lamine, an appropriate reference compound, AH (eq 4),
whose BDE value is already known, and the correspond-
ing radicals.
Ke
\
R2NO-H + A• y
z R2NO• + A-H
(4)
In the present cases, 2,2,6,6-tetramethylpiperidine-1-
hydroxyl and several phenols were used as reference
derivatives. The persistent nitroxide radicals were gener-
ated, at room temperature in degassed solutions of an
appropriate solvent, from the corresponding hydroxy-
lamines by hydrogen exchange with the commercial
nitroxide TEMPO and the less persistent ones by pho-
tolysis with UV light in the presence of about 10% of di-
tert-butylperoxide. Benzene was used as solvent when
possible, while mixtures of benzene and tert-butyl alcohol
or tert-butyl alcohol alone were used with some deriva-
tives sparingly soluble in apolar solvents.
An examination of the measured hyperfine splitting
constants and g-factors, reported in Table 1, shows that
the spectroscopic parameters are strongly dependent on
the nature of the substituent linked to the nitroxide
function. In fact, aN decreases and the g-factor increases
by substituting alkyl groups, linked to the nitrogen atom,
with vinyl or acyl groups. In nitroxides with conjugated
CdX groups, the decrease of the hyperfine splitting
constant at nitrogen is a well-known effect, easily un-
Kinetic studies have also been carried out in order to
measure the absolute rate constants for reactions 1 and
3 and to better clarify the mechanism of the PINO-
catalyzed oxidations. Kinetic EPR was used in the former
case while, to obtain the value of the rate constant for
(5) Lucarini, M.; Pedulli, G. F.; Cipollone, M. J . Org. Chem. 1994,
59, 5063. Lucarini, M.; Pedrielli, P.; Pedulli, G. F.; Cabiddu, S.;
Fattuoni, C. J . Org. Chem. 1996, 61, 9259. Lucarini, M.; Pedulli, G.
F.; Valgimigli, L.; Amorati, R.; Minisci F. J . Org. Chem. 2001, 66,
5456-5462.
(6) Brigati, G.; Lucarini, M.; Mugnaini, V.; Pedulli, G. F. J . Org.
Chem. 2002, 67, 4828-4832.
(7) Lucarini, M.; Pedrielli, P.; Pedulli, G. F.; Valgimigli, L.; Gigmes,
D.; Tordo, P. J . Am. Chem. Soc. 1999, 121, 11546-11553.
(3) McMillen, D. F.; Golden, D. M. Annu. Rev. Phys. Chem. 1982,
33, 493-532.
(4) Mahoney, L. R.; Mendenhall, G. D.; Ingold, K. U. J . Am. Chem.
Soc. 1973, 95, 8610-8614.
1748 J . Org. Chem., Vol. 68, No. 5, 2003

Hydroxylamines as Oxidation Catalysts
derstandable on the basis of the resonance structures
A-D, due to the delocalization of spin density on X when
this is a carbon atom and due to a redistribution of spin
density within the NO group from nitrogen to oxygen
when X is the strongly electronegative oxygen atom.8
Thus, while substitution of an alkyl with an aryl group
induces a decrease of spin density on both nitrogen and
oxygen atoms, substitution with an acyl group produces
a strong decrease of spin density on nitrogen and a small
increase on oxygen, the latter effect being consistent with
the larger g-value measured in acyl nitroxides.8
To measure the BDE values of the O-H bond different
methods of generation were used due to the different
persistence of the nitroxides from 1a -8a and NHPI. In
the case of 1a , 2a , 3a , and 4a , which gave rise to very
persistent radicals, diluted benzene solutions of hydroxy-
lamine ((2-8) × 10-4 M) and of TEMPO (5 × 10-5 M)
were mixed, degassed, and introduced in the EPR cavity
so that the equilibrium constant for the hydrogen atom
exchange (eq 5) could be studied.
F IGURE 1. Experimental EPR spectrum of nitroxide 6b in
benzene at room temperature (above). Below it is shown a
portion of the experimental EPR spectrum obtained by partial
oxidation of a mixture of 6a and BHT and its computer
simulation.
the nitrogen atom in 1a , 2a , and 3a and of both an aryl
and a vinyl group in 4a .
The three acyl hydroxylamines 5a -7a , instead, be-
haved differently. Actually, when mixing solutions of
these hydroxylamines with TEMPO in a wide range of
concentrations, the three-line spectrum of TEMPO re-
mained unchanged and no signals due to the nitroxides
5b, 6b, or 7b were detected. This was a clear indication
that the O-H bond in these acyl hydroxylamines is much
stronger than in TEMPOH or in 1a -4a . Thus, to
measure the BDE values of 5a -7a , reference compounds
having stronger O-H bonds had to be used. By testing
different phenols, 2,6-di-tert-butyl-4-methylphenol (BHT),
characterized by an O-H BDE of 81.0 kcal/mol,5 was
found to be the most suitable one.
TEMPO• + R2NO-H h TEMPO-H + R2NO• (5)
The concentration ratio of the two species was adapted
so that EPR spectra resulting from the superimposition
of lines from both TEMPO and one of the nitroxides 1b-
4b could be simultaneously detected. The relative amounts
of the two radicals were obtained either by numerical
integration of the spectra or by computer simulation,
while the concentrations of the corresponding hydroxyl-
amines were calculated by difference as shown in eq 6,
where [TEMPO]0 and [R2NOH]0 are the initial concentra-
tions of the two species.
The equilibration experiments were done by preparing
solutions of BHT (ca. 0.3 M) and of each hydroxylamine
(ca. 0.03 M) in benzene containing some tert-butyl alcohol
(see Table 1) and di-tert-butyl peroxide as photoinitiator.
The radicals were generated photolytically by a short
flash of filtered UV light. Figure 1 shows, as an example,
the EPR spectra of nitroxide 6b (above) and of its mixture
with the phenoxyl radical from BHT (below). Since these
experiments were performed on concentrated solutions
of the radical precursors, the initial concentration of BHT
and R2NOH could be used in the calculation of Ke (eq 6)
while the relative concentrations of the equilibrating
radical species were determined by both simulation and
numerical integration of the EPR spectra. As expected,
the BDE values for the three acyl nitroxides were found
to be much larger than those of 1a -4a , i.e., 78.5, 79.2,
and 80.2 kcal/mol for 5a , 6a , and 7a , respectively.
When trying to measure the equilibration constant
between 8a , BHT, and the corresponding radicals, com-
plex EPR spectra were obtained that could not be
interpreted due to the formation of unidentified second-
ary radical species. Attempts made by substituting BHT
with 6a were again unsuccessful because, also in this
case, spectra of secondary species were detected.
[R2NO•]([TEMPO]o - [TEMPO])
[TEMPO]([R2NOH]o - [R2NO•])
Ke )
(6)
The equilibrium constant Ke was used to calculate the
free energy of the hydrogen atom transfer reaction from
the hydroxylamine to TEMPO, which was put equal to
the enthalpy of reaction, by assuming that the entropic
contribution is negligible.5 The BDE value of 1a -4a was
then calculated by means of eq 7 by using the known4
BDE value (69.6 kcal/mol) of TEMPOH (i.e., the hydroxyl-
amine precursor of TEMPO).
BDE(R2NO-H) ) BDE(TEMPO-H) - RT ln Ke (7)
The values obtained for 1a , 2a , 3a , and 4a , i.e., 70.6,
71.4, 69.7, and 72.8 kcal/mol, respectively, are close to
those of TEMPOH and of other alkylhydroxylamines4
despite the presence of an aryl group directly bonded to
(8) Aurich, H. G. In The Chemistry of Amino, Nitroso, Nitro
Compounds and of their Derivatives; Patai, S., Ed.; Wiley: Chichester,
1982; Chapter 14.
J . Org. Chem, Vol. 68, No. 5, 2003 1749

Amorati et al.
Finally, the same equilibration procedure was carried
out with mixtures of NHPI and a number of ring-
substituted phenols characterized by O-H bond strengths
larger than 80 kcal/mol. Since NHPI is sparingly soluble
in benzene, tert-butyl alcohol was used as solvent for the
EPR radical equilibration experiments in order to satisfy
one of the conditions requested, i.e., large concentrations
of the two radical precursors.5 When studying mixtures
of NHPI and BHT, only the EPR spectrum of the
phenoxyl radical was observed even by using trace
amounts of the phenol. Since this indicates that the BDE
value of the hydroxylamine is much larger than that of
BHT, other phenols characterized by larger O-H bond
strengths were used. The most suitable one was 3,5-di-
tert-butylphenol, whose BDE value in benzene is 86.6
kcal/mol.5 When photolyzing approximately equimolar
mixtures of this phenol (0.056 M) and of NHPI (0.051
M) containing some di-tert-butyl peroxide, the resulting
phenoxyl and nitroxide radicals were observed in a 1:1
ratio. This means that the BDE value of NHPI is 0.05
kcal/mol lower than that of the reference compound if
entropic contributions are neglected.5 Another phenol
that, when mixed with NHPI (0.012 M), shows the
presence of both radicals is 4-cyanophenol9 (0.94 M). In
this case, the determination of the equilibrium constant
gives a difference between the BDEs of the phenol and
the hydroxylamine of 3.7 kcal/mol. Although the experi-
mental determination of the BDE of 4-cyanophenol has
never been reported, this can be straightforwardly esti-
mated in benzene solution by the substituent additivity
rule10 as 89.0 kcal/mol.6
It is important to emphasize the fact that these
measurements have been carried out in tert-butyl alcohol
where, as in other hydrogen bond acceptor (HBA) sol-
vents, the BDE value of substituted phenols may be
considerably different from that one measured in benzene
due to the solvation of the OH group.11,12 Actually, in the
case of phenols, the BDEs have been found to increase
by ca. 2.2 kcal/mol on passing from benzene to tert-butyl
alcohol when no substituents are present in the ortho
positions, by ca. 1.0 kcal/mol for 2,6-dimethyl substituted
phenols, while in 2,6-di-tert-butylphenols they seem to
be substantially unaffected.11 Thus, we can estimate the
BDE values of the two reference compounds in the
solvent used for the present determinations, as 88.8 and
91.2 kcal/mol for 3,5-di-tert-butylphenol and for 4-cy-
anophenol, respectively, and thus the O-H bond strength
of NHPI in tert-butyl alcohol as 88.1 ( 0.6 kcal/mol.
To estimate the bond dissociation enthalpy in non-HBA
solvents such as benzene, the following considerations
can be made. If NHPI behaves as a phenol without ortho
substituents, the BDE value should be lower than in Me3-
COH by 2.2 kcal/mol (i.e., 85.9 kcal/mol). Since, however,
NHPI contains carbonyl groups at the right distance from
the hydroxyl proton to give an intramolecular hydrogen
bond, it seems more likely that the solvent effect is lower
and similar to that observed in 2-methoxyphenol where
the OH group is intramolecularly hydrogen bonded to the
ortho substituent. A reasonable estimate is that in non-
HBA solvents the BDE value of NHPI should be ca. 1
kcal/mol lower, that is about 87 kcal/mol.
In conclusion, it is clear on the basis of the experimen-
tal results collected that the strength of the O-H bond
in hydroxylamines is strongly dependent on the nature
of the substituents linked to the nitrogen atom. Actually,
while in dialkyl hydroxylamines the O-H BDE value is
ca. 70 kcal/mol, in alkyl acyl hydroxylamines is about 10
kcal/mol larger and in diacyl hydroxylamines is even
larger. To understand the reasons for this behavior, it
should be pointed out that the BDE value is a measure
of the energy difference between the nitroxide radical and
the parent hydroxylamine; thus, any factor inducing a
stabilization of the hydroxylamine or a destabilization
of the nitroxide increases the strength of the O-H bond.
The higher BDE values observed in acyl hydroxylamines
are likely due to the larger resonance stabilization arising
from structures with charge separation that instead do
not contribute to the mesomeric system in dialkyl hy-
droxylamines.
Moreover, as discussed above on the basis of the
computational results reported by Aurich,8 the presence
of the CdO group in acyl nitroxides reduces the impor-
tance of the mesomeric structure (B) due to its electron-
acceptor character, this implying a destabilization of
these species with respect to alkyl nitroxides. Another
reason for the larger BDEs of acyl hydroxylamines is the
likely formation of an intramolecular hydrogen bond that
may stabilize the starting compound, while obviously, it
is absent in the corresponding nitroxide.
It seems of some interest, for prediction purposes, that
the BDE values of the hydroxylamines roughly correlate
with the g-values of the corresponding nitroxide radicals
as shown in Figure 2. An explanation of this behavior
can be found in the fact that the larger values of BDE in
the hydroxylamines and of the g-factors in the corre-
sponding radicals are both the result of the presence of
the electronegative carbonyl substituent which increases
the importance of polar mesomeric structures, as dis-
cussed previously. This correlation may be very useful
for roughly estimating BDE values of hydroxylamines
simply from the measure of the g-factor of the corre-
sponding nitroxides.
Kin etic Exp er im en ts. To measure the rates by which
PINO radicals can abstract a hydrogen atom from a given
substrate (RH, eq 1), and thus the rates of the catalyzed
oxidation by molecular oxygen, we have carried out
kinetic studies by using both EPR spectroscopy to moni-
tor the decay of radical species and a differential pressure
transducer to measure the oxygen consumption in a
closed system. The reactivity of PINO with several
(9) The measured spectroscopic parameters of 4-cyanophenoxyl
radical are: aH(ortho) ) 6.86 G, aH(meta) ) 2.31 G, aN ) 1.38 G, g )
2.0054.
(10) Borges dos Santos, R. M.; Martinho Simo˜es, J . A. J . Phys. Chem.
Ref. Data 1998, 27, 707-739.
(11) Pedrielli, P.; Pedulli, G. F. Gazzetta 1997, 127, 509.
(12) De Heer, M. I.; Korth, H. G.; Mulder, P. J . Org. Chem. 1999,
64, 6969.
1750 J . Org. Chem., Vol. 68, No. 5, 2003

Hydroxylamines as Oxidation Catalysts
F IGURE 2. Correlation among the O-H bond dissociation
enthalpy values of hydroxylamines and the g-factors of the
corresponding nitroxide radicals.
F IGURE 3. Pseudo-first-order rate constants for the decay
of PINO radicals in the presence of different amounts of
hydrogen atom donors in benzene + 10% ACN.
substrates containing an easily abstractable hydrogen
atom has been previously studied in ACN containing
pyridine by Masui and co-workers13 by using UV spec-
troscopy. EPR was used to determine the rate constant
of hydrogen abstraction from RH by PINO, following the
decay of the EPR signals of these radicals in the presence
of increasing amounts of the substrate. PINO was
produced photochemically from NHPI and dicumyl per-
oxide (20% by weigth) in benzene solutions containing
10% acetonitrile, and the investigated compounds RH
were toluene, ethylbenzene, isopropylbenzene (cumene),
benzyl alcohol, cyclohexane, and adamantane. The decay
of PINO followed good first-order kinetics both in the
absence and in the presence of the oxidizable substrate.
In the former case, this is likely due to a fragmentation
at one of the carbonyl carbon-nitrogen bonds, similarly
to what previously reported by Perkins and co-workers
for an other acylnitroxide.14 This behavior differs from
that one described by Masui and co-workers who ob-
served second-order decay when PINO was produced
electrochemically in ACN pyridine mixtures.13 Actually,
we have found that the kinetics of decay of this nitroxide
strongly depends on the time of irradiation and on the
solvent composition. Therefore, all kinetic measurements
were carried out by using freshly prepared samples
shortly irradiated with a flash of light. Under these
conditions, the decay traces were nicely described by eq
7, where kd is the first-order rate constant for self-decay
and kH is the second-order rate constant for hydrogen
abstraction by RH. The factor 2 before kH is required
since the alkyl radical formed by hydrogen abstraction
from RH is expected to subtract a second molecule of the
nitroxide from the solution by a fast combination reaction
to give the adduct PINO-R.
TABLE 2. Absolu te Ra te Con sta n ts, kH, for th e
Hyd r ogen Abstr a ction Rea ction fr om RH by P INO in
Ben zen e + 10% ACN a t 25 °C (Eq 1) a n d by th e
ter t-Bu tylp er oxyl Ra d ica l, kp ,15 a t 30 °Ca
RH
PhCH3
PhCH2CH3
PhCHMe2
PhCH2OH
cyclohexane
adamantane
kH (PINO)/M-1 s-1
kpb/M-1 s-1
kH (PINO)/kp
0.38
2.24
3.25
28.3
0.047
0.047
0.036
0.20
0.22
0.13
0.0034
10.5
11.2
14.8
218
13.8
a
The kd value for the first-order self-decay of PINO in the
absence of hydrocarbons was 0.1 s-1
.
These are the overall rate
b
constants independent of the number of labile hydrogens.
the pseudo-first-order rate constant kEPR ) kd + 2kH[RH]
are plotted in Figure 3 as function of the substrate
concentration and the resulting kH obtained from the
slopes of these plots are shown in Table 2. It should be
pointed out that the presently determined values for
ethylbenzene and benzyl alcohol are similar to those
measured by Masui, i.e., 1.85 and 15.6 M-1 s-1, respec-
tively, in a different solvent and by a different experi-
mental technique.13 In the case of ethylbenzene, mea-
surements were also done by using the perdeuterated
derivative C6D5CD2CD3; the corresponding room-temper-
ature rate constant was 0.22 M-1 s-1, this implying a
kinetic isotope effect of 8.7. Table 2 also reports the rate
constants given in the literature15 for the related hydro-
gen abstraction reaction from RH by tert-butylperoxyl
radicals, kp. It is seen that the reactivity of PINO with
the various substrates is larger than that of tert-butyl-
peroxyl radicals, although the observed difference de-
pends on the nature of the oxidizable substrate being
larger with benzyl alcohol than with the aromatic hy-
drocarbons presumably because of polar effects.2
[PINO]t
ln
) -(kd + 2kH[RH])t
(7)
To better understand the factors determining the rate
of NHPI-catalyzed oxidation of hydrocarbons, we have
also investigated the thermally initiated autoxidation of
cumene, chosen as a model system, by measuring the rate
of oxygen uptake in the presence and in the absence of
NHPI. The autoxidation was followed by monitoring the
oxygen consumption in a closed system with an automatic
[PINO]0
Measurements were carried out by using different
substrate concentrations compatible with a decay lasting
at least 20 s. The experimentally determined values of
(13) Ueda, C.; Noyama, M.; Ohmori, H.; Masui, M. Chem. Pharm.
Bull. 1987, 35, 1372-1377.
(14) Hussain, S. A.; J enkins, T. C.; Perkins, M. J . Tetrahedron Lett.
1977, 36, 3199-3202.
(15) Howard, J . A. In Free Radicals; Kochi, J . K., Ed.; Wiley-
Interscience: New York, 1975; Vol. 2, Chapter 12.
J . Org. Chem, Vol. 68, No. 5, 2003 1751

Amorati et al.
F IGURE 4. Dependence of the oxygen consumption rate, observed during the AMVN (5.0 × 10-3 M) initiated autoxidation of
cumene (0.89 M) at 30 °C in chlorobenzene, on the NHPI concentration (A) and on the content of the ACN in solution (B). The
three plots of part A were obtained at different ACN concentrations (b, 0.5%; ∆, 3%; 0 37.5%), while the data reported in part B
were obtained at an NHPI concentration of 2.5 × 10-4 M.
All these observations can be rationalized on the basis
of the reaction mechanism reported in eqs 8-13, where
the first four reactions describe the hydrocarbon oxida-
tion in the absence of catalyst and the last two represent
the catalytic cycle in the presence of a hydroxylamine.
Ri
9
In
8 R•
(8)
(9)
R• + O2 f ROO•
kp
9
ROO• + RH
8 R• + ROOH
(10)
(11)
(12)
(13)
F IGURE 5. Oxygen consumption rates at 30 °C observed
during the autoxidation of cumene, at increasing concentra-
tion, in chlorobenzene and ACN 0.5% in the presence of AMVN
5.0 × 10-3 M: ∆, without NHPI; O, in the presence of NHPI
2.5 × 10-4 M; 9, difference between the catalyzed and the
noncatalyzed oxygen consumption rates.
2kt
2 ROO• 8 products
kf
R2NO-H + ROO• 98 R2NO• + ROOH
recording gas absorption apparatus, built in our labora-
tory, that uses as detector a commercial differential
pressure transducer.16
kH
R2NO• + RH
9
8 R• + R2NO-H
The autoxidations were carried out at 30 °C in chloro-
benzene solutions containing cumene (0.89 M), 2,2′-
azobis(2,4-dimethylvaleronitrile) (AMVN) as initiator
(0.005 M), and variable amounts of NHPI (0-2.5 × 10-3
M). The oxygen uptake traces showed an initial induction
period of several minutes where the rate of oxidation was
low, then d[O2]/dt increased becoming linear with time.
When plotting this rate of oxygen consumption as func-
tion of the hydroxylamine in solution (see Figure 4), good
linearity was observed for low NHPI concentrations,
while at a higher content of hydroxylamine the rate of
oxidation was lower than expected. To study the effect
of solvent on the rate of reaction, these experiments were
carried out in the presence of increasing amounts of
acetonitrile (ACN), a solvent having good hydrogen bond
accepting properties. It is seen that by increasing the
ACN concentration a large reduction of the rate of
reaction as well as a decrease of the nonlinearity of the
plots are observed. Experiments have also been per-
formed by changing the rate of initiation Ri, i.e., the
AMVN concentration, and the cumene concentration. The
rate of oxidation was found to increase linearly with the
square root of Ri, while the dependence on the oxidizable
substrate concentration is shown in Figure 5.
By solving the simultaneous differential equations
describing the time evolution of the various species,
under the assumption that the steady-state approxima-
tion holds for all free radical species, eq 14 can be
obtained for the rate of oxygen consumption; the first
term to the right describes the noncatalyzed and the
second one the catalyzed oxidation.
1/2
1/2
d[O2]
Ri
Ri
-
) kp[RH]
+ kf[R2NO-H]
(14)
(
)
(
)
dt
2kt
2kt
When using low concentrations of the oxidizable sub-
strate, the only important term is the last one, and
therefore, the rate of oxidation is expected to depend on
(i) the hydroxylamine concentration, (ii) the square root
of the rate of initiation, and (iii) the inverse of the square
root of the rate of termination of the peroxyl radicals from
the substrate. Experimentally, it was found that the first
two predictions are actually verified since -d[O2]/dt
changes linearly with the NHPI concentration, at least
for low contents of the hydroxylamine (see Figure 4), and
with the square root of the initiator concentration.
According to the proposed kinetic model the rate of
oxidation should be independent of the substrate con-
centration. From Figure 5, showing the results of a series
(16) Amorati, R.; Pedulli, G. F.; Valgimigli, L.; Attanasi, O. A.;
Filippone, P.; Fiorucci, C.; Saladino, R. J . Chem. Soc., Perkin Trans. 2
2001, 2142-2146.
1752 J . Org. Chem., Vol. 68, No. 5, 2003

Hydroxylamines as Oxidation Catalysts
of measurements carried out at different cumene con-
centrations, both in the absence and in the presence of
NHPI, it appears that the plot obtained by subtracting
the rate of noncatalyzed to the rate of catalyzed oxidation
does not significantly depend on the cumene concentra-
tion.
than that of dialkyl hydroxylamines, conclusively dem-
onstrates that acyl substituents at the nitrogen atom are
capable of increasing the strength of the hydroxylamine
O-H bond. This is attributed to resonance effects both
in the acyl substituted and in the corresponding nitroxide
radicals as well as to intramolecular hydrogen bonding
in the acyl hydroxylamine.
On the basis of eq 14, the slope of the plot reporting
-d[O2]/dt against [NHPI] will provide kf(Ri/2kt)1/2. Since
both Ri and 2kt can be measured independently, the rate
constant kf for the hydrogen atom abstraction from NHPI
by the peroxyl radical representing the rate-determining
step of the overall reaction can be obtained. However,
due to the low solubility of NHPI in PhCl, all experiments
were performed in the presence of some ACN that, as
shown in Figure 4, reduces the rate of oxidation with
respect to PhCl. The results of a more complete series of
data obtained at different solvent compositions are
reported in Figure 4B. The reduced rate of oxidation
observed at high ACN concentrations is due to the
formation of hydrogen bonding between ACN and the
hydroxyl proton of NHPI and can be analyzed in terms
of a simple model, proposed by Ingold and co-workers,17
based on the assumptions that the HBA solvent, S, and
substrate, R2NO-H, give rise to 1:1 hydrogen-bonded
complex, R2NO-H‚‚‚S, in equilibrium with the hydroxyl-
amine, which is much less reactive toward peroxyl
radicals.
Kinetic studies carried out in order to investigate the
catalytic behavior of NHPI on the oxidation of cumene
are consistent with a simple kinetic model where the rate
determining step is the hydrogen atom abstraction from
the hydroxylamine by the cumylperoxyl radicals (eq 12).
It should, however, be emphasized that this model relies
on the assumption that the rate of hydrogen abstraction
from cumene by the nitroxide radical (PINO) is signifi-
cantly large to allow the steady-state concentration of the
various radical species to be reached within reasonably
short times. With other substrates less reactive than
cumene, the kinetics of the aerobic oxidation might be
more complicated.
Due to the low solubility of NHPI in apolar solvents,
the effect of adding to the reaction medium increasing
concentrations of a hydrogen-bond acceptor solvent, such
as ACN, that can complexate the hydroxylic group of
NHPI, was also investigated. The considerable reduction
of the rate of oxidation, observed in the presence of ACN,
provided additional evidence that the slower step of the
cumene oxidation is the reaction of NHPI with peroxyl
radicals. It seems possible that, with a more lipophilic
diacyl hydroxylamine catalyst, the efficiency of the oxida-
tion could substantially improve.
Exp er im en ta l Section
Ma ter ia ls. Solvents were of the highest grade commercially
available and were used as received. The hydroxylamines 1a ,19
5a ,20 6a ,21 and 7a 22 and the stable nitroxides 2b,23 3b,24 and
4b25 were prepared according to literature procedures, while
8a and NHPI were commercial products. The hydroxylamines
2a (mp ) 159-161 °C from benzene/petroleum ether; IR ν )
3475 and 1600 cm-1), 3a (mp ) 176-177 °C from ligroin; IR
ν ) 3400 and 1600 cm-1), and 4a (mp ) 184-185 °C from
benzene/petroleum ether; IR ν ) 3415 a 3300 cm-1), were
obtained from 2b, 3b, and 4b, respectively, by reduction with
phenylhydrazine according to the method described in ref 19.
EP R Sp ectr a . The EPR spectra were recorded on a Bruker
On the basis of this model, the measured rate constant
for a given solvent composition, kfS will be given by eq
15, where [S] is the concentration of the ACN, kf0 is the
rate constant of the free hydroxylamine, and KS is the
equilibrium constant for the complexation of NHPI by
ACN.
kf0
1 + KS[S]
kfS
)
(15)
Analysis of the data of Figure 4B provides the kf0
)
ESP 300 spectrometer equipped with
a Hewlett-Packard
7.2 × 103 M-1 s-1 and KS ) 1.3 M-1 by using the
measured values of Ri ) 5.5 × 10-9 M s-1 and 2kt )1.6 ×
104 M-1 s-1 in chlorobenzene at 30 °C.18
5350B microwave frequency counter for the determination of
the g-factors, which were corrected with respect to that of
perylene radical cation in concentrated H2SO4 (g ) 2.00258).
The nitroxide radicals were generated photochemically in
deoxygenated benzene or tert-butyl alcohol solutions of the
hydroxylamine under study and di-tert-butyl peroxide (10%
v/v) contained in Suprasil quartz EPR tubes sealed under
nitrogen. The sample was inserted in the cavity of an EPR
spectrometer and photolyzed with the unfiltered light from a
Con clu sion s
The present study provides a rationalization of the
catalytic effect of the phthalimide-N-oxyl radical (PINO)
in the aerobic oxidation of several organic substrates. The
unusually high hydrogen abstracting activity of this
nitroxide radical is justified by the large bond dissociation
enthalpy (BDE) of the O-H bond in the corresponding
hydroxylamine (NHPI), which is ca. 18 kcal/mol larger
than that of dialkyl hydroxylamines. The BDE value of
acyl hydroxylamines, approximately 10 kcal/mol larger
(19) Berti, C.; Colonna, M.; Greci, L.; Marchetti, L. Tetrahedron
1975, 31, 1745-1753.
(20) Do¨pp, D.; Sailer, K. H. Chem. Ber. 1975, 108, 301-313.
(21) Brink, C. P.; Fish, L. L.; Crumbliss, A. L. J . Org. Chem. 1985,
50, 2277.
(22) Alewood, P. F.; Hussian, S. A.; J enkins, T. C.; Perkins, M. J .;
Sharma, A. H.; Siew, N. P. Y.; Ward, P. J . Chem. Soc., Perkin Trans.
1 1978, 1067-1076.
(23) Colonna, M.; Greci, L.; Poloni M. J . Heterocycl. Chem. 1980,
17, 1473-1477.
(17) Avila, D. V.; Ingold, K. U.; Lusztyk, J .; Green, W. H.; Procopio,
D. R. J . Am. Chem. Soc. 1995, 117, 2929-2930.
(18) Lucarini, M.; Pedulli, G. F.; Valgimigli, L. J . Org. Chem. 1998,
63, 4497-4499.
(24) Do¨pp, D.; Greci, L.; Nur-el-Din. Chem. Ber. 1983, 116, 2049-
2057.
(25) Colonna, M.; Greci, L.; Marchetti, L. Gazzetta 1976, 106, 197.
J . Org. Chem, Vol. 68, No. 5, 2003 1753

Amorati et al.
500 W high-pressure mercury lamp. The temperature was
controlled with a standard variable-temperature accessory and
was monitored before and after each run with a copper-
constantan thermocouple.
in the thermostated cavity of a Bruker ESP 300 spectrometer
equipped with a Bruker ER033M field-frequency lock.
PINO was produced photochemically by a short pulse of UV
light from an unfiltered 500 W high-pressure Hg-lamp and
the decay of its ESR spectrum monitored as function of time.
For each experiment, a single time-sweep EPR scan was
recorded with the following settings: microwave power 5 mW,
time constant 4 ms, conversion time 20-164 ms.
Au toxid a tion Exp er im en ts. Autoxidation experiments
were performed in a two-channel oxygen uptake apparatus,
based on a Validyne DP 15 differential pressure transducer,
that has already been described elsewhere.16 The entire
apparatus was immersed in a thermostated bath that ensured
a constant temperature within (0.1 °C.
In a typical experiment, an air-saturated chlorobenzene
solution containing cumene (0.89 M), and variable amounts
of NHPI (0-2.5 × 10-3 M) were equilibrated at 30 °C with a
reference solution containing the same reaction mixture and
a large amount of R-tocopherol (ca. 10-4 M) instead of NHPI.
After equilibration, a concentrated chlorobenzene solution of
the radical initiator 2,2′-azobis(2,4-dimethylvaleronitrile)
(AMVN) (0.005 M) was injected in both the reference and
sample flasks and the oxygen consumption in the sample was
measured, after calibration of the apparatus, from the dif-
ferential pressure recorded with time between the two chan-
nels. This instrumental setting allowed us to have the N2
production and the oxygen consumption derived from the azo-
initiator decomposition already corrected from the measured
reaction rates. Initiation rates, Ri, were determined for each
conditions in preliminary experiments by the inhibitor method
using R-tocopherol as reference antioxidant: Ri ) 2[R-toco-
pherol]/τ. Experiments were also performed by changing the
rate of initiation Ri, i.e., the AMVN concentration, the cumene
concentration, and in the presence of different amounts of
acetonitrile (ACN).
Deter m in a tion of th e O-H BDE Va lu es. The bond
dissociation enthalpies of the O-H bond were obtained from
the equilibrium constants Ke (eq 6) in the case of the reactions
between TEMPO and one of the hydroxylamines 1a -4a , since
all the nitroxide radicals derived from these hydroxylamines
are very long-lived, and thus, the total radical concentration
remains constant during the course of the experiment. These
measurements were carried out by introducing in the EPR
cavity degassed benzene solutions of mixtures of a hydroxy-
lamine ((2-8) × 10-4 M) and TEMPO ((1-5) × 10-5 M) and
by determining the relative concentrations of the two nitrox-
ides either by integration of the EPR signals or by computer
simulation of the spectrum.
In the case of the hydroxylamines 5a -7a and of NHPI, the
reference compound was a phenol whose O-H BDE value was
known. The equilibrating radicals were produced by photo-
lyzing deoxygenated benzene or tert-butyl alcohol solutions
containing the hydroxylamine under investigation (0.1-0.5 M),
an appropriate reference compound (0.1-0.5 M), and di-tert-
butyl peroxide. The molar ratio of the two radicals was
obtained from the EPR spectra and used to determine the
equilibrium constant, Ke (eq 4). The equilibrium constants were
found to be independent of the initial product concentrations
and on the rate of initiation. The initial concentrations of the
two reactants were used for this purpose since these were high
enough to avoid significative consumption during the course
of the experiment. Relative radical concentrations were de-
termined by comparison of the digitized experimental spectra
with computer simulated ones. In these cases, an iterative
least-squares fitting procedure based on the systematic ap-
plication of the Monte Carlo method was performed in order
to obtain the experimental spectral parameters of the two
species including their relative intensities.5
Kin etic Mea su r em en ts by EP R. The rate of hydrogen
abstraction from the investigated hydrocarbons and benzyl
alcohol by the PINO radical was measured at 298 K by kinetic-
EPR. An EPR quartz tube containing a nitrogen-saturated
benzene solution containing 10% acetonitrile, dicumyl peroxide
(20% by weight), and NHPI (10-3 M), and increasing amounts
of the investigated compound, i.e., toluene, ethylbenzene,
isopropylbenzene, benzyl alcohol, and cyclohexane, was placed
Ack n ow led gm en t. Financial support MURST (Re-
search project “Free Radical Processes in Chemistry and
Biology: Synthesis, Mechanisms, Applications”) is grate-
fully acknowledged.
J O026660Z
1754 J . Org. Chem., Vol. 68, No. 5, 2003