Autoxidation of ethylbenzene: The mechanism elucidated

Article abstract of DOI:10.1021/jo070040m

(Chemical Equation Presented) Using a combined experimental and theoretical approach, we elucidated the mechanism of ethylbenzene autoxidation, at about 420 K. The generally accepted literature mechanism indeed fails to explain basic experimental observat

Full text of DOI:10.1021/jo070040m

Autoxidation of Ethylbenzene: The Mechanism Elucidated
,†
‡
†
Ive Hermans,* Jozef Peeters, and Pierre A. Jacobs
2
Centre for Surface Chemistry and Catalysis, Department of Microbial and Molecular Systems (M S),
K.U.LeuVen Kasteelpark Arenberg 23, B-3001 HeVerlee, Belgium, and DiVision for Quantum Chemistry
and Physical Chemistry, Department of Chemistry, K.U.LeuVen, Celestijnenlaan 200F,
B-3001 HeVerlee, Belgium
iVe.hermans@biw.kuleuVen.be
ReceiVed January 8, 2007
Using a combined experimental and theoretical approach, we elucidated the mechanism of ethylbenzene
autoxidation, at about 420 K. The generally accepted literature mechanism indeed fails to explain basic
experimental observations, such as the high ketone to alcohol ratio. The hitherto overlooked propagation
of 1-phenyl-ethylhydroperoxide, the primary chain product, is now unambiguously identified as the source
of acetophenone as well as of 1-phenylethanol via a subsequent activated cage reaction. A similar
mechanism allowed rationalizing of the cyclohexanone and cyclohexanol formation in the autoxidation
of cyclohexane. The primary hydroperoxide product is found to react about 10 times faster than the
arylalkane substrate with the chain carrying peroxyl radicals, whereas in cyclohexane autoxidation, this
reactivity ratio is as high as 55. In combination with a lower efficiency of the above-mentioned cage
reaction, this results in a rather high 1-phenyl-ethylhydroperoxide yield and causes a high ketone/alcohol
ratio. Radicals are shown to be predominantly generated via a concerted bimolecular reaction of the
hydroperoxide with the arylalkane substrate, producing alkyl and hydrated alkoxy free radicals. In this
autoxidation system, no reaction product exhibits a major initiation-enhancing autocatalytic effect, as is
the case with cyclohexanone in cyclohexane autoxidation. As a result, the conversion rate increases less
sharply in time compared to cyclohexane autoxidation. In fact, even some slight inhibition can be observed,
•
due to the formation of chain-terminating HO
2
radicals in the alcohol co-oxidation. At 418 K, the chain
length is estimated to be about 300-500 for conversions up to 10%.
Introduction
nylon-6,6, and the oxidation of p-xylene to terephthalic acid
6
(
30 × 10 tons per year), a building block for poly(ethylene
The selective oxidation of hydrocarbons by molecular oxygen
is of major industrial importance, and improving its efficiency
and selectivity toward the value-added products remains a prime
6
terephthalate). Likewise, cumene hydroperoxide (5 × 10 tons
per year), derived from the oxidation of cumene, serves as
feedstock for both phenol and acetone. Another interesting
autoxidation process is the conversion of ethylbenzene to
1
-3
objective.
Important examples are the autoxidation of cy-
6
clohexane (6 × 10 tons per year), yielding cyclohexanone and
6
ethylbenzene hydroperoxide (5 × 10 tons per year) which is
cyclohexanol, key intermediates in the synthesis of nylon-6 and
used as an epoxidizing agent in the propylene oxide/styrene
process.
*
To whom correspondence should be addressed. Tel: (+32) 16 321648.
Fax: (+32) 16 321998.
In view of the highly complex nature of the autoxidation
mechanisms, the detailed understanding of these processes still
today presents a major scientific challenge. It is widely assumed
that this type of oxidation proceeds via a radical chain
†
Centre for Surface Chemistry and Catalysis, Department of Microbial and
2
Molecular Systems (M S).
‡
Division for Quantum Chemistry and Physical Chemistry, Department of
Chemistry.
(
1) Sheldon, R. A.; Kochi, J. K. In Metal-Catalyzed Oxidations of
Organic Compounds; Academic Press: New York, 1981.
2) Franz, G.; Sheldon, R. A. Oxidation, Ullmann’s Encyclopedia of
Industrial Chemistry; Wiley-VCH: Weinheim, 2000.
(
(3) Bhaduri, S.; Mukesh, D. Homogeneous Catalysis, Mechanisms and
Industrial Applications; John Wiley & Sons: New York, 2000.
1
0.1021/jo070040m CCC: $37.00 © 2007 American Chemical Society
Published on Web 03/16/2007
J. Org. Chem. 2007, 72, 3057-3064
3057

Hermans et al.
mechanism, reactions 1-6, where QdO denotes the respective
rates of propagation and termination, i.e., the chain length, is
very large, resulting in a radical chain mechanism.
1
-3
ketone formed.
Recently, we demonstrated that the mechanism outlined in
reactions 1-6, though widely accepted over many decades, fails
utterly to explain product formation in the autoxidation of
Initiation:
•
•
ROOH f RO + OH
(1)
(2)
(3)
6,7
cyclohexane. Indeed, given the long radical chain length of
about 100, the slow chain termination reaction (6), with a rate
equal to the (equally slow) initiation step, can only account for
a very small fraction of the observed alcohol and ketone
molecules. A combination of state-of-the-art theoretical meth-
odologies and experiments demonstrated that cyclohexyl peroxyl
•
•
RO + RH f ROH + R
•
•
OH + RH f H O + R
2
Propagation:
•
radicals (CyOO ) abstract not only hydrogen atoms from the
•
•
substrate molecule (CyH) but also, and much more rapidly, the
weakly bonded RH-atom from CyOOH. The ratio of the
propagation rate constants for both H-abstraction reactions,
R + O f ROO
(4)
(5)
2
•
•
ROO + RH f ROOH + R
CyOOH CyH
6,7
k
/k , was determined to be as high as 55. The resulting
R-hydroperoxy-alkyl radicals are demonstrated to be unstable
Termination:
•
and to dissociate promptly into OH and QdO (reaction 8),
•
•
8
without any energy barrier. This CyOOH propagation is thus
ROO + ROO f ROH + QdO + O
(6)
2
identified as the fast ketone source that was missing in the
literature so far.
Reaction 1 represents the homolytic dissociation of the weak
O-O bond in the corresponding hydroperoxide, often initially
•
•
1
added in a small quantity to light off the reaction. However,
CyOOH + CyOO f {Cy-RH OOH} +
the efficiency of a unimolecular scission is very low in the liquid
•
CyOOH f Cy-RHdO + OH + CyOOH (8)
•
•
phase as the nascent alkoxy (RO ) and hydroxyl ( OH) radicals
are prone to recombine within their solvent cage to reform
ROOH faster than they can diffuse out of the Franck-
•
The exothermic dissociation of Cy-RH OOH into Cy-RHdO and
OH is rapidly followed by an H-abstraction by the “hot” OH
radical from an ubiquitous RH molecule in the solvent cage
around the nascent products, producing even more heat
reaction 9). The overall exothermicity of more than 60 kcal/
mol causes the formation of a nanosized hot spot, several
hundred degrees kelvin above the bulk temperature, governing
•
•
4
Rabinowitch cage and start off new chains. Nevertheless, many
autoxidation processes show a distinct autocatalytic behavior,
clearly coupled to the growth of the hydroperoxide concentra-
tion. For cyclohexane (CyH), this major issue was settled by
the identification of the bimolecular reaction of the hydroper-
oxide (CyOOH) with the ketone product (Cy-RHdO) as the
9
(
6
the fate of the cage products. Indeed, although reaction 10,
4
predominant initiation mechanism (reaction 7). In this reaction,
the out-of-cage diffusion, features a lower activation barrier,
the subsequent activated-cage reaction (11) accounts for the
majority of the reaction flux (about 70%), due to the high cage
temperature. This cage reaction was identified as the missing
source of CyOH and was shown at the same time to account
•
the OH radical breaking away from the hydroperoxide abstracts
an RH-atom from Cy-RHdO, producing the resonance-stabilized
•
•
ketonyl radical (Cy-RH,-âH dO or Q-R′H dO) and a cyclohexoxy
•
radical (CyO ), hydrogen bonded to H2O. These effects combine
to reduce the energy barrier and prevent geminate recombination
of the product radicals in their solvent cage, thus greatly
enhancing the efficiency of this initiation mechanism.
6,7
for the observed net removal of CyOOH.
•
cage
{
Cy-RHdO + OH + CyOOH + CyH}
f
•
•dO (7)
-R′H
•
cage
CyOOH + Cy-RHdO f CyO ‚‚‚H O + Q
{Cy-RHdO + H
2
O + CyOOH + Cy }
(9)
2
•
• cage
Carbon-centered radicals R , formed via the fast reactions 2
and 3, rapidly react with dioxygen (reaction 4), producing the
{
Cy-RHdO + H O + CyOOH + Cy }
f
2
•
•
Cy-RHdO + H O + CyOOH + Cy (10)
chain carrying peroxyl radicals (ROO ). These radicals can
2
abstract H-atoms from the substrate (reaction 5), yielding ROOH
and new R radicals. Reactions 4 and 5 are chain-propagating
•
cage
•
{Cy-RHdO + H O + CyOOH + Cy }
f
•
2
reactions, which conserve the number of radicals. When the
hydroperoxide product can efficiently initiate new radical chains
via, e.g., reaction 7 for the CyH case, this mechanism should
Cy-RHdO + H O + CyO + CyOH (11)
2
•
The CyO coproduct radicals not only abstract H-atoms from
the substrate (reaction 2) but also undergo ring-opening via â
4
indeed be referred to as autocatalytic. The propagation sequence
is repeated many times before the peroxyl radicals are destroyed
in the termination reaction (6), producing an alcohol (ROH)
and a ketone (QdO). This termination reaction balances the
chain initiation reaction, leading to a fast radical quasisteady
(
6) Hermans, I.; Nguyen, T. L.; Jacobs, P. A.; Peeters, J. Chem. Phys.
Chem. 2005, 6, 637.
7) Hermans, I.; Jacobs, P. A.; Peeters, J. J. Mol. Catal. A: Chem. 2006,
251, 221.
(
5
state. Under normal autoxidation conditions, the ratio of the
(
8) Vereecken, L.; Nguyen, T. L.; Hermans, I.; Peeters, J. Chem. Phys.
Lett. 2004, 393, 432.
4) Hermans, I.; Jacobs, P. A.; Peeters, J. Chem.-Eur. J. 2006, 12, 4229.
(9) At the high temperature of the hot spot, the reactions of the nascent
OH radical with RH and ROOH are nearly equally fast, such that the larger
•
•
•
k
× [ROO ]}. Even at very low conversions, τ is much shorter than the
number of RH molecules, making up the cage wall, as compared to the
single ROOH molecule inside the cage will result in OH reacting mainly
with RH.
•
•
3058 J. Org. Chem., Vol. 72, No. 8, 2007

Ethylbenzene Autoxidation
SCHEME 1. Overall Autoxidation Reaction of Ethylbenzene
SCHEME 2. Proposed Reaction Scheme for the Autoxidation of Ethylbenzenea
a
Q stands for R-RH, in casu Ph-C-CH3.
1
C-C cleavage. Thus, cage reaction 11 was also found to
account for the majority of the ring-opened (acidic) byproducts
in the autoxidation of cyclohexane.7
Cy-RHdO was found to be somewhat less reactive than
•
QdO CyH
CyOH toward CyOO (k
/k
≈ 5), making Cy-RHdO only
,10
6,7,10
a limited source of byproducts,
in contrast to earlier literature
1
-3
An analogous cage reaction can also occur after the H-
abstraction from CyH (reaction 12). As this cage reaction is
not activated, its efficiency is much smaller, so that it contributes
to only 5% of the CyH propagation flux. Still, this minor
reaction channel readily explains the direct formation of (some)
views.
On the other hand, Cy-RHdO was shown to greatly
enhance radical initiation via an efficient bimolecular initiation
4
step (7), as discussed above.
6
The objective of the present work is to elucidate the detailed
mechanism responsible for the autoxidation of ethylbenzene
(Scheme 1), which not only is an important bulk process by
itself but also can serve as a model for the oxidation scheme of
1
1
CyOH from CyH, which thus far could not be rationalized.
•
•
cage
13
arylacetic esters and heterocyclic acetic esters, relevant in fine
CyOO + CyH f {CyOOH + Cy }
f
•
chemical applications.
CyO + CyOH (12)
Similar to CyH autoxidation, the primary chain propagation
•
step, ROO + RH, is expected to involve RH-abstraction from
The peroxyl radical can also abstract the RH-atom from
the secondary -CH2- alkyl substituent moiety. As a working
hypothesis, it is reasonable to propose an ethylbenzene oxidation
mechanism (Scheme 2) along the same lines as that for CyH.
CyOH. Though this reaction is slower than reaction 8, it is still
CyOH CyH
6,7
important as k
/k
≈ 10. The R-hydroxy-alkylperoxyl
•
radical (Cy-RH(OH)OO ), formed upon addition of O2, was
shown to rapidly equilibrate with its products, Cy-RHdO and
HO2 . This equilibrium is strongly shifted toward the products,
•
The specific resonance-stabilized Q OOH radical, resulting from
•
•
12
the RH-abstraction from 1-phenyl-ethylhydroperoxide by ROO ,
6
was earlier shown to decompose in the absence of an energy
thus constituting another, though minor, ketone source. More
•
8
barrier into the corresponding ketone and OH.
importantly, the very fast, diffusion-controlled radical-termina-
tion reaction (13) was argued to cause a substantial decrease in
the peroxyl radical concentration, such that the CyOH coau-
toxidation significantly slows the overall kinetics.
Results and Discussion
1
. Experimental Observations. Figure 1a shows the ethyl-
benzene to product conversion in time. For safety reasons, the
conversion was limited to less than 10% to avoid a build-up of
peroxides. A short induction period, during which radicals need
to be formed via unfavorable pathways, can be observed. Such
an induction stage can be bypassed if a small amount of ROOH
•
•
CyOO + HO f CyOOH + O
(13)
2
2
(
10) Hermans, I.; Jacobs, P. A.; Peeters, J. Chem.-Eur. J. 2007, 13,
54.
11) Berezin, I. V.; Denisov, E. T.; Emanuel, N. M. The Oxidation of
Cyclohexane; Pergamon Press: New York, 1996.
12) Hermans, I.; M u¨ ller, J.-F.; Nguyen, T. L.; Jacobs, P. A.; Peeters, J.
J. Phys. Chem. A 2005, 109, 4303.
7
(
(
(13) See, e.g.: Wentzel, B. B.; Donners, M. P. J.; Alsters, P. L.; Feiters,
M. C.; Nolte, R. J. M. Tetrahedron 2000, 56, 7797.
J. Org. Chem, Vol. 72, No. 8, 2007 3059

Hermans et al.
FIGURE 2. Product distribution against the sum of products in the
418 K autoxidation of ethylbenzene: 1-phenyl-ethyl-hydroperoxide (2);
acetophenone (×); and 1-phenylethanol (b).
FIGURE 1. Ethylbenzene conversion in time at 418 K: (a) pure
ethylbenzene, (b) ethylbenzene with 0.7 mol % of 1-phenyl-ethyl-
hydroperoxide initially added.
is initially added (Figure 1b). In contrast to cyclohexane
autoxidation,4 the conversion increases nearly linearly, rather
than exponentially, pointing to the less-pronounced autocatalytic
nature of ethylbenzene autoxidation.
often overlooked nonterminating peroxyl self-reaction,
ROO + ROO f RO + RO + O2.
,6
•
•
•
•
17
In Figure 2, the product distribution is plotted against the
sum of products. Only products arising from the oxidation of
secondary C-H bonds in the substrate could be observed
(Scheme 1). This plot clearly identifies ROOH as the only
primary reaction product, whereas both ROH and QdO are
secondary products and must therefore originate from ROOH.
A striking observation is the high peroxide selectivity (e.g., 83%
after 6.3% conversion), compared to the cyclohexyl hydroper-
oxide selectivity in the autoxidation of cyclohexane (only 25%
Obviously, at low conversions, the overall oxidation rate,
dΣ[P]/dt, is equal to the RH conversion rate by reaction 5
(eq 1).
RH
•
dΣ[P]/dt ) k × [ROO ] × [RH]
(eq 1)
The rate constant for RH propagation, kRH, can be estimated
as follows. At the validated B3LYP/6-311++G(df,pd)//B3LYP/
6
after 6.3% conversion).
2
. Input from Computational Chemistry. The higher
6
6
-31G(d,p) level of theory, the barrier for H-abstraction from
1
-phenyl-ethylhydroperoxide yield from ethylbenzene should
•
ethylbenzene by CH3OO radicals was calculated to be 12.06
kcal/mol. Combining this activation barrier with the measured
ROOH RH
be ascribed to a significantly reduced k
to the k
/k ratio compared
CyOOH CyH
/k
ratio of 55 for cyclohexane. This ratio
RH
-1 -1 14
k
at 303 K, 1.3 M s , allows the calculation of a pre-
corresponds to the RH-abstraction reactions from the (aryl)-
alkane substrate and the corresponding hydroperoxide by the
dominant chain-propagating peroxyl radicals, i.e., 1-phenyl-
ethylperoxyl and cyclohexylperoxyl, respectively. This is ra-
tionalized in Figure 3 by comparing the calculated activation
barrier for the abstraction of H-atoms from different alkanes
8
-1
exponential (TST) rate factor per H-atom of 3.25 × 10 M
-
1 15
RH
8
-1 -1
s . It follows that k is equal to 6.5 × 10 M
s
× exp-
-
1
(
-12.06 kcal mol /RT). On the basis of the experimental
RH
-1 -1
dΣ[P]/dt and k at 418 K of 320 M s , one can estimate
the 1-phenyl-ethylperoxyl radical concentration, using expres-
sion 1. Thus, with 0.7 mol % of ROOH initially added and
after 1.0% RH conversion, i.e., for [ROOH] ) 0.13 M, one
•
and their corresponding hydroperoxides by the CH3OO radical.
It can be observed that the reactivity difference between the
hydroperoxide and the alkane decreases when the substrate
becomes more reactive. For substrates with secondary C-H
bonds (C-H BDE0 ) 96.6 kcal/mol), the difference in the
abstraction barrier between the alkane and the peroxide is
predicted to be 5.5 kcal/mol, whereas for substrates featuring
much weaker bonds (such as ethylbenzene, C-H BDE0 ) 80.5
kcal/mol), this difference decreases to only 1.9 kcal/mol.
Therefore, it is predicted that the RH-abstraction of 1-phenyl-
ethylhydroperoxide is only about 5 times as fast as the
H-abstraction from the parent alkane. The barrier for abstraction
of the RH-atom from 1-phenylethanol is calculated to be 2.5
•
-8
finds [ROO ] ≈1.9 × 10 M. Combining this number with
term
the rate constant for chain termination, k (418 K) ) 1.8 ×
8
-1 -1
1
0 M
s
(vide infra), a chain length of 385 is obtained.
This large number demonstrates indeed that chain termination
reaction 6 cannot be responsible for a sizable fraction of the
1
reaction products as often assumed. Moreover, according to
the classical mechanism (reactions 1-6), the expected QdO/
ROH ratio should be close to 1/2, whereas an experimental value
of approximately 2 is observed. Indeed, according to this simple
scheme, ketone is produced exclusively in the termination
reaction whereas an almost equal amount of additional alcohol
originates from reaction 2, following the initiation step (see also
below). In fact, even more alcohol could be produced via an
ROH RH
kcal/mol lower than that from the RH substrate. A k /k
ratio of about 10 seems reasonable, implying that the observed
QdO/ROH ratio of 2 cannot be ascribed to a fast subsequent
oxidation of the alcohol.
(14) Howard, J. A. Reactions of Organic Peroxyl Radicals in Organic
Solvents. In Peroxyl Radicals, The Chemistry of Free Radicals; Alfassi,
Z. B., Ed.; John Wiley & Sons: Chichester West Sussex, 1997; p 283.
(
15) This value is in good agreement with the pre-factor per H-atom,
(17) Lightfoot, P. D.; Cox, R. A.; Crowley, J. N.; Destriau, M.; Hayman,
G. D.; Jenkin, M. E.; Moortgat, G. K.; Zabel, F. Atmos. EnViron. 1992,
26A, 1805.
(18) Luo, Y.-R. Handbook of Bond Dissociation Energies in Organic
Compounds; CRC Press: Boca Raton, 2003.
perH
•
A
, derived from the known rate constant for CH3OO + C3H8 f CH3-
• -1 -1 16
OOH + iso-C3H7 at 400 K, i.e., 0.96 M
of 16.1 kcal/mol, i.e., A
s , and the calculated barrier
perH
) 3.0 × 10 M s-1
8
-1
.
(16) Tsang, W. J. Phys. Chem. Ref. Data 1988, 17.
3060 J. Org. Chem., Vol. 72, No. 8, 2007

Ethylbenzene Autoxidation
FIGURE 3. Computed barriers for the abstraction of H-atoms from
different alkanes (a) and from the corresponding hydroperoxides (b)
as a function of the alkane C-H bond strength,18 reduced to 0 K.
FIGURE 4. Plot of the experimental [ROH]/[ROOH] ratio vs the [Qd
O]/[ROOH] ratio, validating eq 2: intercept ) 0.006 ( 0.002,
slope ) 0.425 ( 0.02, correlation coefficient ) 0.993.
sec
Substrates from left to right: ethylbenzene, toluene, propane (H ),
ethane, and methane.
benzaldehyde, or its oxidation product benzoic acid, in the
2
2
reaction mixture.
. Reaction Analysis. Following the mechanism outlined in
•
SCHEME 3. Pathways for Reaction of the RO Alkoxy
Radical
3
Scheme 2, but neglecting the integrated effect of ROH coau-
toxidation for a conversion up to 10%, eq 2 can be derived
7
based on a stoichiometric ROOH balance.
[
ROH]
b
d
a
[QdO]
≈ a(1 + e) + (1 + e)
(eq 2)
[
ROOH]
[ROOH]
In Figure 4, the [ROH]/[ROOH] ratio is plotted vs the [Qd
O]/[ROOH] ratio, demonstrating that at up to nearly 10%
conversion the linear relation proposed in eq 2 is observed
indeed. From the intercept of the plot in Figure 4, and using e
≈
0.9, it follows that fractions a and b are equal to 0.997 and
.003, respectively. Apparently, the cage fraction is nearly
0
•
Next, the fate of the RO radicals is addressed. These alkoxy
negligible, in contrast to the case of cyclohexane autoxidation
where the value of b is equal to 0.05. This difference can be
readily understood when comparing the computed activation
energy for this cage reaction: for cyclohexane and ethylbenzene,
this barrier is 6.8 vs 8.4 kcal/mol, respectively (B3LYP/6-31G-
(d,p) level). Obviously, the higher barrier must be attributed to
radicals can abstract H-atoms from the alkane substrate or
decompose via â-scission (Scheme 3). The rate of H-abstraction
fraction e in Scheme 3) can be estimated as follows. Given
the 298 K rate constant of the CyO + CyH and (CH3)3CO +
(
•
•
5
5
-1 -1
CyH abstraction reactions of 8.0 × 10 and 9.2 × 10 M s ,
respectively,1
9,20
and the calculated barrier of (CH3)2CHO +
•
resonance stabilization in the R radical, lowering its reactivity.
•
CyH of 4.7 kcal/mol, one can estimate a pre-exponential (TST)
From the slope of Figure 4, values can be obtained for the
fractions c and d of 0.78 and 0.22, respectively. The efficiency
of the activated cage reaction is also much lower than that for
cyclohexane, due to this difference in activation barriers.
As the ratio of branching ratios, (d/c)/(b/a) ≈ 95, should equal
exp[∆E/R × ∆(1/T)], with ∆E being the difference between
the barrier of the activated cage reaction (≈8.4 kcal/mol) and
the barrier for diffusive separation (≈1.5 kcal/mol) and ∆(1/T)
referring to the difference between the hot spot T and the bulk
T ()418 K), one can estimate an effective hot spot T of
≈900 ( 200 K. This is in line with the rough estimation
stemming from the 65 kcal/mol of heat released in the fast
propagation process.
sec
8
-1 -1
rate factor per H -atom of 2.0 × 10 M s , in line with the
•
analogous pre-factor per H-atom for abstractions by ROO
8
-1 -1
radicals (i.e., 3.25 × 10 M s , vide supra). Note that the
influence of the alkyl group in the oxy radical on the abstraction
reaction from a given substrate is very small. Combining this
pre-factor with the barrier for H-abstraction from PhCH2CH3
•
•
by (CH3)2CHO , a model for RO , i.e., 3.2 kcal/mol, shows a
6
-1 -1
rate constant of 8.5 × 10 M
s
at 418 K. The pseudo-first-
•
order rate constant for RO removal via H-abstraction can thus
7
-1
be estimated at ≈7 × 10 s . On the other hand, the â-cleavage
•
of PhCH(O )-CH3 (fraction f in Scheme 3) proceeds at a rate
13
-1
of k(T) ) 5.3 × 10 × exp(-12.7 kcal/mol/RT) s (TST
calculation based on a B3LYP/6-31G(d,p)-PES, known to give
Within the frame of the new reaction mechanism, eq 3 can
be derived on the basis of a kinetic analysis of the alcohol and
21
reliable results for this type of reaction ). Clearly, for this
specific RO radical, the H-abstraction channel will be fa-
•
7
ketone formation rates.
vored: e ≈ 0.9 and f ≈ 0.1, as witnessed by the absence of
RH
[
ROOH] d[ROH]
k
(19) Druliner, J. D.; Krusic, P. J.; Lehr, G. F.; Tolman, C. A. J. Org.
×
≈ (1 + e) × b ×
d[QdO]
+
ROOH
[RH]
Chem. 1985, 50 (26), 5843.
k
(
(
20) Weber, M.; Fischer, H. J. Am. Chem. Soc. 1999, 121 (32), 7381.
21) Peeters, J.; Fantechi, G.; Vereecken, L. J. Atmos. Chem. 2004, 48,
[
ROOH]
(
1 + e) × d ×
(eq 3)
5
9.
[RH]
J. Org. Chem, Vol. 72, No. 8, 2007 3061

Hermans et al.
reactivity of the hydroperoxide, and (ii) the lower efficiency of
the activated cage reaction which is a net sink of hydroperoxide
(see Scheme 2).
4
. Formation of New Radicals During the Chain Initiation.
ROOH RH
As the k
/k ratio is now known, eq 1 can be corrected by
taking into account the additional RH consumption, subsequent
to the hydroperoxide propagation (see Scheme 2), leading to
eq 5.
RH
•
dΣ[P]/dt ) {(1 + b × e) × k × [ROO ] × [RH]} +
ROOH
•
{
(1 + d × e) × k
× [ROO ] × [ROOH]} (eq 5)
This expression allows for a more accurate estimation of the
peroxyl radical concentration, which can ultimately be used to
evaluate k /k , via the radical quasisteady state (eq 6).
FIGURE 5. Validation of eq 4 by plotting the experimental [ROOH]/
init term
5
[
RH] × d[ROH]/d[QdO] product ratio vs the [ROOH]/[RH] ratio:
-4
intercept ) (5.8 ( 5.0) × 10 , slope ) 0.52 ( 0.01, correlation
term
• 2
init
coefficient ) 0.999.
k
× [ROO ] ) k × [ROOH]
(eq 6)
Whereas kinit/kterm increases by an order of magnitude during
autoxidation of cyclohexane (Figure 6a), a moderate but
monotone decrease of this ratio can be observed during
ethylbenzene autoxidation (Figure 6b). The pronounced auto-
catalytic effect in the CyH autoxidation was earlier ascribed to
occurrence of reaction 7: cyclohexanone is efficiently assisting
4
the CyOOH molecules to initiate new radical chains. Due to
the very fast increase in cyclohexanone during CyH autoxida-
init term
tion, k /k also increases sharply. Nevertheless, at higher CyH
conversions, the enhancement of the initiation was found to be
•
counteracted by HO2 formation, as witnessed by the sigmoidal
init term
4
shape of the k /k
ratio (Figure 6a). As the RH-atoms of
acetophenone are bound more strongly than those in cyclohex-
anone, the former compound is much less efficient in assisting
the hydroperoxide cleavage. Actually, a steady decrease in
init term
FIGURE 6. Measured kinit/kterm ratio vs the sum of products during
the autoxidation of cyclohexane (a) and ethylbenzene (b) at 418 K.
k /k can be observed during the ethylbenzene autoxidation.
This remarkable effect was also observed by Korthals Altes et
al., when examining the ethylbenzene oxidation rate, derived
0
.5
In Figure 5, the experimentally determined left-hand side of
eq 3, corrected for ROH co-oxidation, is plotted against the
from the O2 consumption, as a function of [ROOH] (i.e.,
•
23
∝ [ROO ], eq 6). To explain this enigmatic behavior, this early
•
•
[ROOH]/[RH] ratio. Again, a linear relation is observed.
study proposed additional termination channels of OH and RO
Combining the intercept of the plot in Figure 5 with the value
radicals with each other and with the main chain propagator
of b derived above, we can estimate the kROOH/k ratio at
RH
•
ROO as becoming important at higher conversions. However,
•
1
0 ( 10, in agreement with the first principles prediction used
even when taking into account the hitherto overlooked OH and
ROOH RH
•
•
above. The k
relative ROOH conversion, viz., {Σ[P] - [ROOH]}/0.5-
[ROOH](0) + [ROOH](t)}, where the rate-averaged [ROOH]
0 to t) is approximated by 0.5{[ROOH](0) + [ROOH](t)}, and
/k ratio can also be extracted from the
RO sources via the ROOH propagation (vide supra), the OH
•
and RO concentrations remain very small throughout the
{
(
reaction due to their fast H-abstraction reactions with RH, i.e.,
9
6
-1 -1
k ) 2 × 10 and 8.5 × 10 M s , respectively, at 418 K.
•
the relative RH conversion, viz., Σ[P]/[RH], taking into account
that ROOH is only consumed in the activated cage reaction,
viz., channel d in Scheme 2 (eq 4).
Thus, as an example, for [ROOH] ) 0.13 M and [ROO ] )
-
8
•
•
1.9 × 10 M (vide supra), [ OH] and [RO ] are calculated to
-
16
-13
be only 3 × 10 and 1.3 × 10 M, respectively. Therefore,
even diffusion-controlled termination channels involving these
radicals will remain entirely negligible compared to the ROO•
self-reaction (6).
ROOH RH
k
/k
≈
{
Σ[P] - [ROOH]}/0.5 × {[ROOH](0) + [ROOH](t)}
1
× d
A far more reasonable explanation for the observed decrease
Σ[P]/[RH]
init term
•
in the k /k
ratio is the formation of HO2 radicals in the
(eq 4)
12
ROH RH
co-oxidation of ROH. Indeed, given the predicted k /k
ratio of 10 (vide supra) and the observed ROH/RH ratio of 0.008
The average value obtained for kROOH/kRH via eq 4 is equal to
•
at 10% conversion, the HO2 production can be estimated to be
2
0 ( 10, in line with the analysis above. From the slope of the
•
a fraction 0.08 of the ROO production. Taking into account
plot in Figure 5, a value for fraction d of 0.27 is obtained, also
in good agreement with the value derived from eq 2.
•
the high cross-termination rate constant for HO2 radicals with
The reason for the much higher peroxide yield in the
ethylbenzene autoxidation than in the cyclohexane oxidation
can thus be attributed to (i) the significantly lower relative
•
•
(
22) Cleavage of the Ph-CH(O )CH3 bond, yielding Ph and CH3CHO,
faces a barrier of 18.5 kcal/mol and can therefore be ruled out completely.
(23) Korthals Altes, F. W.; van den Berg, P. J. Recueil 1966, 85, 538.
3062 J. Org. Chem., Vol. 72, No. 8, 2007

Ethylbenzene Autoxidation
•
9
-1 -1 24
ROO (≈2 × 10 M s ) vs the mutual termination rate
this reaction is in good agreement with its predicted energy
barrier and the analogous reaction in CyH autoxidation.
The improved peroxyl radical concentration, obtained via eq
5, can also be used to get a more accurate value of the chain
length. Due to the nearly linear increase of the ROOH
concentration as a function of Σ[P] (Figure 2), the ROO•
concentration also increases because the rate of chain initiation
is proportional to [ROOH] × [RH] (vide supra). As a conse-
quence, the chain length is found to decrease upon increasing
conversion, although it remains very high (Figure 8).
•
8
-1 -1
constant for ROO radicals (1.8 × 10 M s , vide infra),
•
this small amount of HO2 can indeed enhance the apparent
termination rate constant by a factor of about 2. This corresponds
init term
to the observed decrease in the k /k ratio in Figure 6b. From
init term
the k /k
ratio, extrapolated to zero conversion, i.e., 2.3 ×
-
15
•
1
0
M, and the mutual termination rate constant for ROO
init
radicals, k can be evaluated. It is reasonable to adopt a
termination rate constant for 1-phenyl-ethylperoxyl radicals
similar to that for phenyl-methylperoxyl ()benzylperoxyl)
8
-1 -1
radicals, i.e., 1.8 × 10 M s , at 418 K from recent gas-
phase rate coefficient measurements2
5,26
and the known tem-
perature dependence of the nontermination/termination branch-
1
7
•
ing ratio. Indeed, for fast self-reacting ROO radicals such as
benzylperoxyl, an additional substituent on the RC (i.e., a methyl
group) appears to affect the rate constant only marginally. Using
term
8
-1 -1
init
a value for k
of 1.8 × 10 M s , a pseudo-first-order k
of 4 × 10 s-1 is found; this value is in line with the apparent
pseudo-first-order rate constant for the “pure” CyOOH initiation
in the cyclohexane autoxidation.4
-7
In our detailed studies on the initiation of cyclohexane
autoxidation, we concluded that even the so-called “pure”
initiation is in fact a bimolecular reaction of CyO-OH with
-
7
-1 -1
Cy-H, reaction 14, with a rate constant of 1.0 × 10
M
s
at 418 K and an experimental Arrhenius activation energy of
FIGURE 7. Potential energy surface of the homolytic dissociation of
ROOH, compared with the bimolecular reaction (15).
about 25 kcal/mol, which is significantly lower than the CyO-
OH BDE of ≈40 kcal/mol.
4
•
•
CyO-OH + CyH f CyO ‚‚‚H O + Cy
(14)
2
Likewise, an experimental rate constant for reaction 15 of
-
8
-1 -1
5
× 10
M
s
can be derived, which is two times smaller
than that for the analogous reaction with CyH. This result should
be attributed to the larger number of H -atoms in cyclohexane
sec
compared to ethylbenzene, compensated by a lower calculated
energy barrier for reaction 15, i.e., 26.6 vs 28.1 kcal/mol for
reaction 14 (B3LYP/6-311++G(df,pd)//B3LYP/6-31G(d,p) level).
•
•
RO-OH + R-H f RO ‚‚‚H O + R
(15)
2
FIGURE 8. Evaluated chain length vs the sum of products during the
418 K autoxidation of ethylbenzene.
Figure 7 compares the potential energy surface of this
bimolecular initiation mechanism (15) with the hitherto assumed
homolytic cleavage, reaction 1. Clearly, the newly proposed
reaction is energetically favorable and proceeds through a loose
TS (O‚‚‚O length ≈ 1.98 Å). An analogous initiation reaction
between two ROOH molecules can be neglected, based on the
Conclusions
In this contribution, the mechanism of ethylbenzene autoxi-
dation was studied in detail, combining complementary experi-
mental and theoretical methodologies. The hitherto widely
accepted reaction scheme was unable to explain several
experimental observations pertaining to the basics of the
mechanism. This literature mechanism attributed the formation
of ketone exclusively to the slow termination reaction between
two chain carrying peroxyl radicals; this reaction, together with
the initiation that should proceed at an equal rate, would result
in an expected alcohol/ketone ratio of at least 2 because the
nonterminating mutual peroxyl reaction to form the alkoxy
radicals plus oxygen must lead to additional alcohol. Experi-
mentally, however, an alcohol/ketone ratio of 0.5 is observed
over a wide conversion range. More importantly, the radical
chain length of this process is so long (>300) that the
termination and initiation reactions cannot possibly contribute
significantly to the product flux; obviously, it is the much faster
chain-propagation steps of the peroxyl radicals that determine
2
5.1 kcal/mol energy barrier and the low ROOH/RH ratio.
Whereas the main fate of the nascent RO and OH radicals
•
•
in the case of the homolytic dissociation will be in-cage
recombination, the sharp potential energy drop after the TS of
the bimolecular initiation mechanism (15) forces a translational
separation of the two radical fragments. Moreover, in-cage
combination is also reduced by the 3.2 kcal/mol strong H-bond
•
between H2O and RO , screening the radical site in the alkoxy
radical. Therefore, in ethylbenzene autoxidation, radicals should
predominantly be formed via reaction 15. The derived rate of
(
24) Rowley, D. M.; Lesclaux, R.; Lightfoot, P. D.; Nozi e` re, B.;
Wallington, T. J.; Hurley, M. D. J. Phys. Chem. 1992, 96, 4889.
25) Noziere, B.; Lesclaux, R.; Hurley, M. D.; Dearth, M. A.; Wallington,
T. J. J. Phys. Chem. 1994, 98, 2864.
26) Dib, G. E.; Chakir, A.; Roth, E.; Brion, J.; Daumont, D. J. Phys.
Chem. A 2006, 110, 7848.
(
(
J. Org. Chem, Vol. 72, No. 8, 2007 3063

Hermans et al.
the product makeup. Indeed, the fairly high observed yields of
ketone and alcohol and the experimental alcohol/ketone ratio
of 0.5 can be readily explained in the frame of a new reaction
scheme. This revised mechanism attributes the formation of
acetophenone to the RH-abstraction from 1-phenyl-ethylhydro-
peroxide by the chain-propagating 1-phenyl-ethylperoxyl radical
and formation of 1-phenylethanol to a subsequent activated cage
reaction, analogous to the formation of cyclohexanone and
cyclohexanol in the autoxidation of cyclohexane. The ethyl-
benzene-derived hydroperoxide is indeed about 10 times more
reactive toward the chain-propagating peroxyl radical than the
arylalkane substrate itself. Interestingly, as a comparison,
cyclohexylhydroperoxide reacts even about 55 times faster than
its cyclohexane parent with cyclohexylperoxyl radicals. A
straightforward explanation for this difference in behavior was
found in the convergence of the reactivity of hydroperoxides
and their parent alkanes as the alkane C-H bond becomes
weaker. Additionally, it was discovered that, due to a higher
activation barrier, the efficiency of the above-mentioned cage
reaction is much lower than that in the cyclohexane autoxidation.
The combined effects afford a higher hydroperoxide yield and
a reversed ketone/alcohol ratio compared to cyclohexane
autoxidation. Beside elucidating the formation mechanism of
the major reaction products, also the detailed mechanism of
chain initiation was identified and kinetically quantified.
Radicals are shown to be predominantly generated via a
concerted bimolecular reaction of the hydroperoxide with the
in the cyclohexane autoxidation. As a result, the conversion rate
increases less sharply in time than in cyclohexane autoxidation.
init term
In fact, an observed decrease in the k /k
ratio, attributed to
•
enhanced termination by HO2 radicals arising from alcohol
coautoxidation, indicates a slight inhibition of the radical chain
mechanism.
This study shows the generic character of the newly proposed
autoxidation mechanism of (substituted) alkanes. It enables us
to explain quantitatively the product distribution for two entirely
different substrates in a straightforward way and is fully backed
up by quantum chemical calculations. The detailed knowledge
of the chemistry at issue constitutes a great leap forward in
understanding the reaction and possibly in further optimization
of this important process.
Experimental and Theoretical Methods
The autoxidation of ethylbenzene (50 mL, p.a.) was studied at
418 K in a 100 mL stainless steel high-pressure Parr reactor, stirred
at 500 rpm. Before heating the reactor, it was pressurized with 2.76
MPa of dioxygen (99.99% purity). Prior to each experiment, the
reactor was passivated with a saturated sodium pyrophosphate (p.a.)
solution.6 The products were quantified by GC-FID, after the
addition of an external standard (1-heptanol, p.a.) via a double
injection: trimethyl phosphine (1 M in THF) was added to one of
the samples to reduce the peroxide product to the alcohol. From
the quantified alcohol content before and after reduction, both the
alcohol and peroxide yields can be determined. The injection-port
temperature of the GC was set at 250 °C. Peak areas were converted
to concentrations by means of the specific sensitivities determined
by direct calibrations.
•
•
arylalkane substrate, producing RO ‚‚‚H2O + R . In this
autoxidation system, there is no reaction product with a major
initiation-enhancing autocatalytic effect, as cyclohexanone has
Quantum chemical (QC) calculations were carried out with the
GAUSSIAN03 program.27 At the DFT level, the Becke three-
parameter hybrid exchange functional was used, combined with
the Lee-Yang-Parr nonlocal correlation functional B3LYP-
DFT.28 Unless stated otherwise, the B3LYP/6-311++G(d,p)//
B3LYP/6-31G(d,p) level of theory was used, which was shown
earlier to agree within 0.5 kcal/mol with state-of-the-art computa-
(27) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin,
K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.;
Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.;
Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.;
Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li,
X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.;
Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.;
Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.;
Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich,
S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck,
A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul,
A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.;
Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham,
M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.;
Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian
6
tional levels for H-abstractions by peroxyl radicals. For other
reactions, the accuracy of the calculated barriers is estimated to be
(
2 kcal/mol.29 Rate coefficients of predominant reaction steps in
the mechanism were evaluated by means of transition-state theory
(TST), using the QC-generated energy and ro-vibrational param-
30
eters. For some specific reaction types, TST pre-factors validated
by experimental data were adopted from the literature.
Acknowledgment. This work was performed in the frame-
work of an IAP project (federal government), IDECAT (Euro-
pean government), CECAT (K.U.Leuven) and two GOA
projects (K.U.Leuven). I.H. thanks the FWO for a research
position.
0
3, revision C.02; Gaussian, Inc.: Wallingford, CT, 2004.
28) (a) Becke, A. D. J. Chem. Phys. 1992, 96, 2115. (b) Becke, A. D.
J. Chem. Phys. 1992, 97, 9173. (c) Becke, A. D. J. Chem. Phys. 1993, 98,
(
5
648. (d) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785.
29) See, e.g.: (a) Young, D. C. Computational Chemistry; John Wiley
Sons: New York, 2001. (b) Koch, W.; Holthausen, M. C. A Chemist’s
(
&
Supporting Information Available: Geometries and energetic
and vibrational data of interesting structures are provided. This
materialisavailablefreeofchargeviatheInternetathttp://pubs.acs.org.
Guide to Density Functional Theory, 2nd ed.; Wiley-VCH: Weinheim,
001.
30) (a) Eyring, H. J. Chem. Phys. 1934, 3, 107. (b) Steinfeld, J. I.;
2
(
Francisco, J. S.; Hase, W. L. Chemical Kinetics and Dynamics; Prentice
Hall: New Jersey, 1989.
JO070040M
3064 J. Org. Chem., Vol. 72, No. 8, 2007