Activation parameters as mechanistic probes in the TAML iron(V)-oxo oxidations of hydrocarbons

Article abstract of DOI:10.1002/chem.201405024

The results of low-temperature investigations of the oxidations of 9,10-dihydroanthracene, cumene, ethylbenzene, [D₁₀]ethylbenzene, cyclooctane, and cyclohexane by an iron(V)-oxo TAML complex (2; see Figure 1) are presented, including product identification and determination of the second-order rate constants k₂ in the range 233-243 K and the activation parameters (ΔH^? and ΔS^?). Statistically normalized k₂ values (log k₂′) correlate linearly with the C-H bond dissociation energies D_C-H, but ΔH^? does not. The point for 9,10-dihydroanthracene for the ΔH^? vs. D_C-H correlation lies markedly off a common straight line of best fit for all other hydrocarbons, suggesting it proceeds via an alternate mechanism than the rate-limiting C-H bond homolysis promoted by 2 . Contribution from an electron-transfer pathway may be substantial for 9,10-dihydroanthracene. Low-temperature kinetic measurements with ethylbenzene and [D₁₀]ethylbenzene reveal a kinetic isotope effect of 26, indicating tunneling. The tunnel effect is drastically reduced at 0 °C and above, although it is an important feature of the reactivity of TAML activators at lower temperatures. The diiron(IV) μ-oxo dimer that is often a common component of the reaction medium involving 2 also oxidizes 9,10-dihydroanthracene, although its reactivity is three orders of magnitude lower than that of 2.

Full text of DOI:10.1002/chem.201405024

DOI: 10.1002/chem.201405024
Full Paper
&
Reaction Mechanisms
Activation Parameters as Mechanistic Probes in the TAML Iron(V)–
Oxo Oxidations of Hydrocarbons
Soumen Kundu, Jasper Van Kirk Thompson, Longzhu Q. Shen, Matthew R. Mills,
Emile L. Bominaar, Alexander D. Ryabov,* and Terrence J. Collins*[a]
Abstract: The results of low-temperature investigations of
the oxidations of 9,10-dihydroanthracene, cumene, ethylben-
zene, [D10]ethylbenzene, cyclooctane, and cyclohexane by an
iron(V)–oxo TAML complex (2; see Figure 1) are presented,
including product identification and determination of the
second-order rate constants k2 in the range 233–243 K and
the activation parameters (DHꢀ and DSꢀ). Statistically
normalized k2 values (log k2’) correlate linearly with the CꢀH
bond dissociation energies DCꢀH, but DHꢀ does not. The
point for 9,10-dihydroanthracene for the DHꢀ vs. DCꢀH corre-
lation lies markedly off a common straight line of best fit for
all other hydrocarbons, suggesting it proceeds via an alter-
nate mechanism than the rate-limiting CꢀH bond homolysis
promoted by 2. Contribution from an electron-transfer
pathway may be substantial for 9,10-dihydroanthracene.
Low-temperature kinetic measurements with ethylbenzene
and [D10]ethylbenzene reveal a kinetic isotope effect of 26,
indicating tunneling. The tunnel effect is drastically reduced
at 08C and above, although it is an important feature of the
reactivity of TAML activators at lower temperatures. The
diiron(IV) m-oxo dimer that is often a common component
of the reaction medium involving 2 also oxidizes 9,10-
dihydroanthracene, although its reactivity is three orders of
magnitude lower than that of 2.
the iron(V)–oxo species toward hydrocarbons.[11] While this
project was being conducted, as fully reported here, related
studies were carried out by two other research groups. Firstly,
Sen Gupta and co-workers conducted a kinetic investigation of
the oxidation of hydrocarbon CꢀH bonds by a room tempera-
ture stable, Generation V TAML iron(V)–oxo complex 3.[12] Then,
Nam et al. reported on the kinetics and mechanism of oxida-
tion of hydrocarbons by 2,[13] duplicating the exact study that
we have been engaged in for several years. The approaches
employed, the reaction conditions selected, the features of the
experimental work emphasized, and the foci of the two investi-
gations have turned out to be different such that in this report
we will attempt to integrate the combined findings into an op-
timal mechanistic assessment. The Nam group’s study and this
work were performed at different temperatures—the Nam
group selected 08C whereas we collected our data in the tem-
perature range from ꢀ40 to ꢀ308C. This discrepancy impacts
the observed chemistry. The stability of 2, which spontaneous-
ly undergoes reduction, FeV!FeIV,[10] is considerably higher at
the lower temperatures, such that kinetic data collected by
monitoring the FeV!FeIV transformation at 08C for slow CꢀH
oxidations could be affected by the spontaneous reduction;
we examine this difference in this work. Rate measurements at
different temperatures as reported herein allow for the deter-
mination of the activation parameters, DHꢀ and DSꢀ. These
parameters are particularly useful for revealing intimate
mechanistic details and for integrating the results from the
different research groups. Herein, we establish the importance
of tunneling in the CꢀH bond oxidation by 2, which
significantly controls the processes at lower temperatures.
Introduction
TAML activators such as 1 (Figure 1) were designed at Carnegie
Mellon University (CMU) to be functional replicas of peroxidase
and cytochrome P450 enzymes.[1–4] In iron TAML systems, mul-
tiple FeIV derivatives in aqueous solutions[5,6] and iron(V)–oxo
species in organic nitriles[7] are readily accessible. The TAML
iron(V)–oxo complexes resemble the active sites of peroxidase
and cytochrome P450 oxidase enzymes.[8,9] The reactivity stud-
ies that have since followed have all been founded on detailed
analyses of the spectroscopic properties of TAML iron(III), -(IV),
and -(V) species, especially when generated from 1 by meta-
chloroperoxybenzoic acid (mCPBA) in acetonitrile at ꢀ408C.[10]
The elementary reactions, FeIII!FeIV, FeIV!FeV, FeV!FeIV, and
the FeIII +FeV comproportionation, were first mapped quantita-
tively as prerequisites to substrate reactivity studies. Then the
first substrate oxidations were focused on the conversion of
organic sulfides to the corresponding sulfoxides. The expected
high reactivity of TAML iron(V)–oxo complexes was confirmed
and substrate-controlled electron and oxygen-atom transfer
mechanisms were revealed.[10] Following this work three years
ago, we began a thorough investigation of the reactivity of
[a] Dr. S. Kundu, J. V. K. Thompson, Dr. L. Q. Shen, M. R. Mills,
Prof. E. L. Bominaar, Prof. A. D. Ryabov, Prof. T. J. Collins
Department of Chemistry, Carnegie Mellon University
4400 Fifth Avenue, Pittsburgh, PA, 15213 (USA)
E-mail: ryabov@andrew.cmu.edu
tc1u@andrew.cmu.edu
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201405024.
Chem. Eur. J. 2015, 21, 1803 – 1810
1803
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Figure 1. Structures of TAML complexes mentioned in this work: iron(III) TAML 1, iron(V)–oxo complex 2, synthesized from 1 and used in this study, and its
previously investigated analogue 3.[12] The diiron(IV)-m-oxo dimer 4 is the comproportionation product of 1 and 2.
Results and Discussion
Products
Scheme 1. Non-rebound mechanism in hydrocarbon oxidation by
Complex 2 was generated (ꢁ95% yield) in acetonitrile at
iron(V)–oxo TAML.
ꢀ408C by treating 1 with one equivalent of mCPBA.[10] Using
just one equivalent is important because, in this case, all the
oxidant should be consumed by the oxidation of FeIII to FeV.
This eliminates potential complications while collecting kinetic
data by following the reduction of FeV. From this point of view,
our approach is more advantageous than that used by Nam
and co-workers, who employed three equivalents of mCPBA,
reporting its complete consumption.[13] When generated at
ꢀ408C, 2 is notably more stable than at 08C, but still reacts
with the hydrocarbons 9,10-dihydroanthracene (DHA), cumene,
ethylbenzene, cyclooctane, and cyclohexane at rates that are
convenient for kinetic measurements. The hydrocarbons were
oxidized to the corresponding alcohols and ketones. Cumene
gave a mixture of 2-phenyl-2-propanol and acetophenone.
Cyclic alkanes were oxidized to the corresponding cyclic alco-
hols and ketones. Ethylbenzene afforded 1-phenylethanol
(83ꢂ2% yield) and acetophenone (16ꢂ4% yield), based on
a FeV to FeIV transformation. Interestingly, the Nam group also
reported the formation of styrene and markedly higher
amounts of acetophenone,[13] which may be associated with
the use of three equivalents of mCPBA in the generation of 2.
When ethylbenzene was oxidized in the presence of H218O
(0.2% v/v), no measurable incorporation of 18O was detected in
either the products 1-phenylethanol or acetophenone, despite
the fact that 2 is known to undergo a rapid oxygen exchange
with water.[7] When Nam and co-workers used H218O, the prod-
uct 1-phenylethanol underwent 15% 18O incorporation[13]
which maybe a consequence of different temperatures em-
ployed. However, the lack of 18O in products generated in the
presence of H218O in our study suggests rapid coupling of O2
with alkyl radicals produced upon H-atom abstraction by 2;
indeed, both the studies by Sen Gupta and co-workers and
Nam and co-workers found differing product distributions
when O2 was eliminated. There was a case where the reaction
product did not contain oxygen; 9,10-dihydroanthracene af-
forded a mixture of anthracene (32% yield) and anthrone
(65% yield). In general, the product profiles in our study agree
with the mechanism shown in Scheme 1, in which an alkyl
radical and FeIVꢀOH are produced in the rate-limiting step with
the rebound process not occurring, consistent with the
conclusions of the Sen Gupta and Nam studies.[12,13]
Kinetic data
Naturally, the oxidation of hydrocarbons by 2 occurs more
slowly at ꢀ408C than at 08C. Therefore the kinetic data could
be collected by accurately measuring initial rates of the disap-
pearance of 2 at 630 nm (see the Supporting Information for
details). In contrast, Nam and co-workers applied an exponen-
tial fitting of the entire kinetic curves making no correction for
the spontaneous reduction of FeV to FeIV.[13] Ethylbenzene was
chosen as a representative substrate. Figure 2A shows the
spectral changes that occur on its addition to the solution of 2
at ꢀ408C. The hydrocarbon reduces FeV, forming FeIV, which
was characterized by Nam and co-workers as the monomeric
[FeIV(TAML)(mCBA)]ꢀ (mCBA=meta-chlorobenzoic acid) species
(5) with the isosbestic points at 560 and 713 nm. The data in
Figure 2A reveal similar spectral changes.
In this study, the initial rates were acquired both by observ-
ing the disappearance of FeV and the generation of FeIV (see
the Supporting Information). There is perfect agreement be-
tween the two methods (Figure 2B and Supporting Informa-
tion). The range of concentrations employed for ethylbenzene
and 2 were, respectively, (0.49–30.0)ꢁ10ꢀ3 m (Figure 3A) and
(0.44–4.4)ꢁ10ꢀ4 m (Figure 3B). The linear plots found in both
cases support the rate law of Eq. (1), in agreement with the
findings of Nam and co-workers.[13]
ꢀd½2ꢃ=dt ¼ k2 ½2ꢃ½hydrocarbonꢃ
ð1Þ
Similar plots were observed for reactions of 2 with all hydro-
carbons studied. The rate constants k2 for ethylbenzene (as
well as for other hydrocarbons including [D10]ethylbenzene)
were measured at different temperatures ranging from ꢀ40 to
ꢀ308C (see the Supporting Information). Although a broader
temperature range is preferred for calculating the activation
Chem. Eur. J. 2015, 21, 1803 – 1810
www.chemeurj.org
1804
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Figure 2. A) Spectral changes associated with reduction of 2 by ethyl-
benzene. Data points were collected every 15 s and spectra are shown for
90 s intervals. The inset shows the decrease in absorbance at l=630 nm;
B) matching consumption of FeV and formation of FeIV (see the Supporting
Information). Conditions: [2]=2.16ꢁ10ꢀ4 m, [ethylbenzene]=5ꢁ10ꢀ3 m,
[H2O]=0.2% (v/v), CH3CN, ꢀ408C.
Figure 3. Initial rates of reduction of 2 by ethylbenzene as a function of
[ethylbenzene] (A) and [2] (B). Conditions: [2]=2ꢁ10ꢀ4 m (A), [ethyl-
benzene]=4.9ꢁ10ꢀ3 m (B), 0.2% H2O (v/v), CH3CN, ꢀ408C.
and DSꢀ) obtained from the linear ln(k2/T) vs. Tꢀ1 plots (see the
Supporting Information) are collected in Table 1. The products
k2 [hydrocarbon] at ꢀ408C were all larger than the rate con-
parameters, practical constraints (see the Supporting Informa-
tion) limited the temperature range used to 108C. The rate
constants k2, calculated from plots such as that in Figure 3A,
together with the corresponding activation parameters (DHꢀ
stant for self-decay kdecay of 2 (1.0ꢁ10ꢀ5 sꢀ1 [10]
under the same
)
conditions, to ensure no competition from self-decay. Also, the
Table 1. Rate constants k2 (measured directly and extrapolated) and activation parameters for reactions of hydrocarbons with 2 in acetonitrile at different
temperatures.
1
2
k2
3
4
5
k2
6
k2
7
k2
8
k2
9
Substrate
DHꢀ
DSꢀ
DCꢀH
[Mꢀ1 sꢀ1
]
[kJmolꢀ1
]
[JKꢀ1 molꢀ1
]
[Mꢀ1 sꢀ1
(08C)[a]
]
[Mꢀ1 sꢀ1
]
[Mꢀ1 sꢀ1
]
[Mꢀ1 sꢀ1
]
[kJmolꢀ1
]
(ꢀ408C)
230(2)[b]
0.091(1)
(08C)[13]
(258C)[a]
(258C)[12]
9,10-dihydroanthracene
cumene
ethylbenzene
[D10]ethylbenzene
cyclooctane
19ꢂ2
ꢀ117 ꢂ7
ꢀ202ꢂ2
ꢀ153ꢂ6
ꢀ119 ꢂ15
ꢀ100ꢂ10
ꢀ70ꢂ40
1150
0.31
1.14
0.13
1.05
–
2400
0.55
2.94
0.54
4.34
0.25
–
326
354
364
–
385
416
14.1ꢂ0.5
25ꢂ1
0.22
0.45
0.04
0.72
0.37
0.79
0.28
–
0.145(1)
0.00566(4)
0.0457(4)
2.59(4)ꢁ10ꢀ4
39ꢂ3
39ꢂ3
–
cyclohexane
60ꢂ10
0.0295
0.022
[a] Values extrapolated from the data reported herein; [b] 38ꢂ1mꢀ1 sꢀ1 for [D4]DHA.
Chem. Eur. J. 2015, 21, 1803 – 1810
www.chemeurj.org
1805
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
rate vs. hydrocarbon concentration had no significant positive
intercept (e.g., Figure 3A), which contributions from the self-
decay of 2 would necessarily produce.
more slowly.[12] The comparison of columns 7 and 8 in Table 1
shows clearly that ethylbenzene and cyclohexane are oxidized
by 2 more than an order of magnitude faster than by 3. Rather
surprisingly, cumene reacts with 2 and 3 at similar rates. This
might result from the lower enthalpy of activation DHꢀ for 2,
which for cumene equals just 14 kJmolꢀ1, making the rate of
Reactivity comparisons
reaction for
temperature.
2 with cumene remarkably insensitive to
The availability of enthalpies of activation DHꢀ (Table 1) pres-
ents an opportunity to compare the values of the rate con-
stants k2 obtained by Nam et al. at 08C[13] with those reported
in this work at lower temperatures and to assess the compara-
tive reactivity of complexes 2 and 3[12] where the data for 3
was obtained at 258C. Therefore the rate constants k2 in
Table 1 were extrapolated to both 0 and 258C. The corre-
sponding k2 values are shown in Table 1 where the k2 values
obtained by Nam et al. and Sen Gupta et al. are displayed as
well. For all but one hydrocarbon (cyclohexane), our extrapo-
lated rate constants are reasonably consistent with the results
reported by Nam and co-workers.[13] The fact that in all cases
their rate constants, obtained by measuring the disappearance
of FeV, are lower than ours may indicate that the expected re-
generation of FeV from the FeIV product by the excess mCPBA
for generating 2 in situ is indeed occurring in their study. It is
also possible that some or all of the small differences may
arise from the tunneling effect (see below). The rate constants
for cumene, ethylbenzene, and cyclooctane oxidation are simi-
lar (cf. columns 5 and 6 in Table 1). However, our value of k2
for cyclohexane is lower by more than an order of magnitude.
This discrepancy is to be expected based on our knowledge of
the instability of 2 at 08C; the spontaneous decomposition will
intrude to measurably reduce [FeV] when the oxidation of
less-reactive molecules such as cyclohexane is being followed
to artificially increase the observed rate constant.
Activation Parameters
Part 1
The rate constants for CꢀH bond activation by high-valent iron
complexes have been extensively reported,[4,14–16] but they are
much less frequently accompanied by the activation parame-
ters DHꢀ and DSꢀ,[17,18] bringing additional value to the data in
Table 1. The negative entropies of activation DSꢀ lie in the
range of ꢀ200 to ꢀ70 JKꢀ1 molꢀ1, supporting a bimolecular
rate-limiting step.[19] Of particular interest are the low enthal-
pies of activation DHꢀ of 14, 19, and 25 kJmolꢀ1 for cumene,
9,10-dihydroanthracene, and ethylbenzene, respectively, for
which the dissociation energies for the corresponding CꢀH
bonds are 354, 326, and 364 kJmolꢀ1, respectively. Low activa-
tion enthalpies such as these tend to be found in enzymatic
processes with highly ordered transition states and are accom-
panied by large and negative activation entropies. Here, the
activation entropies are typical of biological reactions. Appa-
rently, the TAML iron(V)–oxo complexes facilitate hydrogen
atom transfer from a general population of reactant/substrate
orientations.
As found by Nam and co-workers at 08C, we found a linear
relationship at all studied temperatures (ꢀ40 to ꢀ308C) be-
tween log k2’ and DCꢀH, here k2’ is a variant of k2 statistically
corrected by the number of the weakest CꢀH bonds
(Figure 4A), supporting rate-limiting CꢀH bond cleavage. It is
worth noting that such plots are not linear in all systems. For
It is also interesting to compare the reactivity of complexes
2 and 3 with respect to hydrocarbon CꢀH bonds. Complex 3 is
considerably more stable than 2 in MeCN under ambient con-
ditions, that is, the spontaneous FeV!FeIV reduction occurs
Figure 4. Log k2’ at different temperatures (A) and DHꢀ (B) plotted against CꢀH bond dissociation energies DCꢀH of hydrocarbons for reactions with 2. Rate
constants k2’ are statistically corrected values of k2 by the number of the weakest CꢀH bonds. See text for details.
Chem. Eur. J. 2015, 21, 1803 – 1810
www.chemeurj.org
1806
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
example, a curved log k vs. DCꢀH plot is also known, where the
interpretation invoked a potential relationship between the
DHꢀ and TDSꢀ terms.[20]
Part 2
In Figure 4A, we compare in the standard manner the rate
constants and the bond dissociation energies (DCꢀH), where
DCꢀH is the enthalpy of CꢀH homolytic cleavage at 298 K.[21] The
plots of log k2’ vs. DCꢀH are linear in all three studies with TAML
activators,[12,13] including this one, as is true for the over-
whelming number of cases where oxidizing iron complexes
have been studied in this way, giving support to rate-limiting
CꢀH bond cleavage.[4] By correlating log k2’ with DCꢀH
,
non-identical thermodynamic functions (Gibbs free energy of
activation vs. enthalpy of homolysis) are being compared
where, in fact, one has the option to correlate the enthalpy of
activation against the enthalpy of homolysis. The correspond-
ing DHꢀ vs. DCꢀH correlation is presented in Figure 4B. In con-
trast with the log k2’ vs. DCꢀH plot, the experimental DHꢀ value
for 9,10-dihydroanthracene is significantly larger than what
would be obtained by linear extrapolation of the DHꢀ vs. DCꢀH
dependence where linearity holds (r2 =0.99) for the other hy-
drocarbons studied in this work. The relatively large positive
slope of 0.72 found for the r2 =0.99 line supports rate-limiting
homolytic CꢀH bond activation with significant CꢀH bond
elongation in the product-like late transition state. The data in
Figure 4B signal that the mechanism for 9,10-dihydroanthra-
cene may be different from the other hydrocarbons. Note that
the lines in Figure 4A do not detect this difference. Moreover,
for a common mechanism to apply across the entire hydrocar-
bon series, the value of DHꢀ for 9,10-dihydroanthracene re-
quired to maintain the linearity of the DHꢀ vs. DCꢀH plot would
be approximately ꢀ4 kJmolꢀ1 (indicated by the arrow in Fig-
ure 4B). A negative value of DHꢀ for a bond that is stable
under ambient conditions is not physically realistic. The mea-
sured value of 19 kJmolꢀ1 is physically realistic. It is also worth
noting that this study supports a previous assertion by
Fertinger, Franke, and van Eldik in a related study that “a close
correlation between bond strength and reaction rate…no
longer exists”.[22]
Figure 5. Correlation between DHꢀ and DSꢀ for the CꢀH bond activation of
hydrocarbons (including [D10]ethylbenzene, C6D5C2D5) by 2. Data are taken
from Table 1.
mechanism for all studied hydrocarbons other than 9,10-di-
hydroanthracene.[23]
Kinetic isotope effect and tunneling effects
Nam and co-workers reported a value of 11 for the kinetic iso-
tope effect (KIE; kH2/k2D for ethylbenzene) when measured at
08C,[13] which does not differ much from the “classical value” of
the KIE (ca. 8) estimated on the basis of the zero-point energy
(ZPE) difference between CꢀH and CꢀD bonds of 4.77 kJmolꢀ1
(1.14 kcalmolꢀ1).[19] However, in the absence of variable tem-
perature studies, Nam and co-workers were unable to establish
a convincing case for the possibility of a tunneling effect. At
the lowest temperature (ꢀ408C) of study in this work, the KIE
increased to 26 for the ethylbenzene/[D10]ethylbenzene pair
(Table 1). At ꢀ34 and ꢀ308C, the KIE values are 22 and 18,
respectively. All the values obtained at lower temperatures are
much higher than both the KIE obtained at 08C and the classi-
cal value, thereby indicating a tunneling effect in the cleavage
of the benzylic CꢀH bond by 2. Using our data from Table 1
and extrapolating the rate constants k2 for ethylbenzene and
[D10]ethylbenzene to 08C, the KIE was found to be
approximately 9, in acceptable agreement with the value of
11 reported by Nam and co-workers.[13]
Further evidence that 9,10-dihydroanthracene follows a dif-
ferent pathway is presented in Figure 5, which demonstrates
an isokinetic plot in which all hydrocarbons (including
[D10]ethylbenzene) other than 9,10-dihydroanthracene show
a linear correlation between DHꢀ and DSꢀ. Once again, the
9,10-dihydroanthracene point lies significantly off the straight
line. In other words, the enthalpy–entropy compensation
effect[23] holds for all studied hydrocarbons other than 9,10-di-
hydroanthracene. The slope of the DHꢀ vs. DSꢀ plot gives the
isokinetic temperature Tiso, at which the rate constants for all
reagents (hydrocarbons in this case) are the same;
The role of tunneling effects in the cleaving of CꢀH bonds
by dissimilar agents was, for example, augmented and re-
viewed several decades ago by Bell.[24] There are three key cri-
teria for tunneling effects in terms of the Arrhenius equation
(k=AeꢀE /RT): i) The KIE kCꢀH/kCꢀD itself; ii) the ratio ACꢀD/ACꢀH
,
a
where ACꢀD/ACꢀH @1,[24] and iii) a difference in activation energy
(DEa) between CꢀH and CꢀD substrates greater than the ZPE
difference of 4.77 kJmolꢀ1 [25]
Our data for ethylbenzene and
.
T
iso =310ꢂ30 K. Importantly, this value is distinctly higher than
[D10]ethylbenzene give an ACꢀD/ACꢀH ratio of 62 and DEa of
14.23 kJmolꢀ1 supporting the presence of the tunneling effect.
In a forthcoming publication, we will detail a theoretical
analysis of the tunneling effect in the TAML systems.
the temperature range over which the k2 values were mea-
sured (233–243 K), validating that the isokinetic relationship is
not an artifact and supporting again a common reaction
Chem. Eur. J. 2015, 21, 1803 – 1810
www.chemeurj.org
1807
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
General mechanistic comments
transfer from the hydrocarbon than when it is hydrogen
atom transfer to 2, as the latter mechanism should require
higher ordering of the transition state for CꢀH cleavage.
The electron and proton transfer should occur in a concerted
manner, which is consistent with both the KIE value of 6 and
the data in Figures 4B and 5. The sp3 CꢀH bonds of DHA are
significantly more acidic (pKa =30.3[32]) than those of cumene
(pKa =41)[33] or cyclohexane (pKa =52),[34] which allows to con-
sider the proton transfer for DHA. The substrate pKa[29] and the
basicity of the oxo ligand[30] can drastically affect the mecha-
nism. The low pKa of DHA[29] and the basicity of 2 may favor
the electron/proton transfer mechanism (Scheme 2). The elec-
tron should move from the HOMO of DHA,[35] supported by its
lowest ionization potential (8.3 eV)[36] compared to those for
cumene and ethylbenzene (8.6[37] and 8.7[38] eV, respectively).
Thus, although it is known to be particularly difficult to
differentiate between the mechanisms mentioned above (see,
for example, the recent publication by Usharani and
co-workers),[39] the experimental data available for DHA is not
inconsistent with the pathway in Scheme 2.
The results described herein support the non-rebound mecha-
nism suggested recently by Nam and co-workers (Scheme 1).[13]
The rebound mechanism is typical of catalysis by cytochrome
P450 where the CꢀH bond homolysis involving iron–oxo reac-
tive intermediates occurs within the confines of the protein
pocket that helps to organize the structure of the interaction
and protect the radical formed from dissociation and sub-
sequent reaction with O2.[26] A protecting pocket is absent in
catalysis by 2 and the radical intermediate once produced has
a higher probability of being trapped by O2 than in the case of
cytochrome P450. A new source of evidence for rate-limiting
hydrogen atom abstraction by 2 is present in Figure 4B. The
linear correlation between DHꢀ and DCꢀH for all hydrocarbons
except 9,10-dihydroanthracene with a slope of 0.72, which is
close to 1, is consistent with a significant CꢀH bond elongation
in the transition state.
Pathway for 9,10-dihydroanthracene oxidation
Comparative reactivity of iron–oxo TAML species: FeIV vs.
FeV
This study has unexpectedly revealed the singular nature of
the rate-limiting step for 9,10-dihydroanthracene oxidation. It
is worth restating that the evidence for this is not observed in
Figure 4A, but instead appears in Figures 4B and 5. Figure 4B
is particularly convincing, because it suggests a negative value
of DHꢀ for 9,10-dihydroanthracene, if it were to share
a common mechanism with the other hydrocarbons studied.
Several mechanistic options for the 9,10-dihydroanthracene ox-
idation can be reasonably suggested: i) Proton-coupled elec-
tron transfer; ii) hydride transfer; iii) a distinct stepwise proton
transfer followed by electron transfer. These pathways have
been discussed in the literature,[27–30] and minor structural var-
iations in the reacting partners and reaction conditions are
known to cause distinct mechanistic alterations. Our current
data do not allow us to suggest with confidence a detailed
nature of the rate-limiting step for 9,10-dihydroanthracene oxi-
dation. The KIE value of 6 for the pair DHA/[D4]DHA at ꢀ408C
points to a different mechanism, because it is drastically lower
than that for ethylbenzene. It should, however, be mentioned
that KIE may depend on the CꢀH bond dissociation energy
Finally, since FeIV and FeV TAML species are intricately involved
in the TAML activator catalytic cycle,[5] the reactivity of the m-
oxo-(FeIV)2 dimer 4 was studied for comparison with that of 2.
Since iron(IV) species are less reactive than iron(V) species,[10]
9,10-dihydroanthracene, as the most reactive hydrocarbon,
was selected. Its addition to a solution of 4 (produced by
adding 0.5 equivalents of mCPBA to 1) in acetonitrile at ꢀ408C
resulted in the oxidation of 9,10-dihydroanthracene to anthra-
cene and anthrone (reaction time=1 h, GC-MS data). The reac-
tivity of 4 was estimated by measuring the initial rates of its
decay at 750 nm, as was done previously for sulfide oxida-
tion.[10] The initial rates varied linearly with the concentration
of 4 (Figure 6A) but leveled off with increasing amounts of
9,10-dihydroanthracene (Figure 6B). This saturation is in agree-
ment with the reversible formation of an adduct between
9,10-dihydroanthracene (DHA) and 4 (K) which collapses into
products (k): 4+DHAÐ{4,DHA} K’, {4,DHA}!products (k). Such
a mechanism leads to Equation (2) for the rate of consumption
of 4.
and a lower value of KIE has been reported for a substrate
[31]
with lower DCꢀH
.
At present, we can only speculate that,
d½4ꢃ k K ½4ꢃ ½DHAꢃ
since the hydrogen atom transfer (HAT) mechanism is unlikely
for 9,10-dihydroanthracene, there might be a significant contri-
bution from an electron transfer pathway in which an electron
moves from the electron-rich hydrocarbon to 2 (Scheme 2).
The highest reactivity found for 9,10-dihydroanthracene arises
from the rather positive entropic term (Table 1). Less entropy
should be lost when the rate-limiting step is an electron
ð2Þ
ꢀ
¼
dt
1 þ K ½DHAꢃ
The data in Figure 6B was fitted to Equation (2) to obtain
K=(1.07ꢂ0.03)ꢁ103 mꢀ1 and k=(7.51ꢂ0.06)ꢁ10ꢀ5 sꢀ1. At low
[DHA], ꢀd[4]/dt=kK[4][DHA] and, for comparison of FeIV and
FeV species, the product kK=(8.0ꢂ0.2)ꢁ10ꢀ2 mꢀ1 sꢀ1 should be
compared with the second-order rate constant k2 for 9,10-dihy-
droanthracene (Table 1). Thus, 2
is more reactive than
4 by
a factor of 2.9ꢁ103 (Table 1). The
reactivity gap is slightly lower
for 9,10-dihydroanthracene than
that for the methyl phenyl sul-
Scheme 2. Tentative mechanism of oxidation of 9,10-dihydroanthracene by 2.
Chem. Eur. J. 2015, 21, 1803 – 1810
www.chemeurj.org
1808
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
there is a contribution from an electron transfer pathway in
the rate-limiting step.
Experimental Section
Materials
TAML activator 1 was synthesized as previously described.[40]
9,10-Dihydroanthracene was obtained from M. P. Biomedicals and
recrystallized from ethanol. 9,9,10,10-[D4]Dihydroanthracene was
synthesized and purified following an established method.[41]
1H NMR spectroscopy and GC-MS indicated >95% deuteration.
Ethylbenzene (Fluka, GC standard) and cumene (Fluka, analytical
standard) were used as received. Cyclooctane (Aldrich, ꢁ99%) and
cyclohexane (Sigma–Aldrich, ꢁ99.7%) were purified by distillation
under reduced pressure. Anthrone and cyclooctanol (both M. P. Bi-
omedicals), anthracene (Aldrich), 1-phenylethanol and cycloocta-
none (both Aldrich, 98%), acetophenone (Fluka, GC standard), 2-
phenyl-2-propanol (Fluka, ꢁ98%), cyclohexanone (Acros, 99.8%
extra pure), and cyclohexanol (Sigma–Aldrich, 99% reagent plus)
were used as received. meta-Chloroperoxybenzoic acid (mCPBA,
Acros Organics, 70–75%) was purified by an established method.[42]
Acetonitrile, water, and ethanol (all HPLC grade, Fisher Scientific)
were used as received. Isotopically enriched H2O18 (97% 18O) was
supplied by Aldrich.
Methods
UV/Vis spectroscopic studies were carried out using an
Agilent 8453 diode array spectrophotometer. Low-temperature
spectral studies were performed using a liquid-nitrogen-cooled
cryostat set-up from UNISOKU Scientific Instruments, Japan. All
spectral data processing was performed using the SigmaPlot 10.0
software package. A Bruker 300 MHz instrument was used for
1H NMR spectroscopy. A Thermo Finnigan gas chromatograph (GC)
for detecting the oxidation products was equipped with a Restek
Rxi-XLB (30 m, 0.25 mm ID, 0.25 mm) column, a Trace DSQ mass
spectrometer, a programmable temperature vaporization (PTV) in-
jector, and a COMBI PAL autosampler (LEAP technologies, CTC Ana-
lytics). Ultra pure grade helium (Penn Oxygen & Supply Company,
PA, USA) at a flow rate of 1 mLminꢀ1 was used as a carrier gas. All
GC-MS experiments were performed in the standard electron-
impact (EI) ionization mode. Products of hydrocarbon oxidation
were confirmed by matching their mass spectra in a reference
library and by comparing retention times of products and stand-
ards. For quantifying 1-phenylethanol and acetophenone formed
from ethylbenzene, temperatures of an injection port and a transfer
line were 250 and 3008C, respectively. The ion source was kept at
2508C. The mass spectrometer was turned on after 4 min during
each analysis. The electron ionization mode was applied (70 eV).
The sample (1 mL) was injected in a splitless mode. The chromato-
graphic oven temperature was programmed in the following way:
Held at 508C for 5 min, ramped at 158C minꢀ1 to 2508C, ramped
at 208C minꢀ1 to 3008C, and held at 3008C for 5 min. Under these
conditions, the retention times of 1-phenylethanol and aceto-
phenone were 10.27 and 10.48 min, respectively. GC-MS data was
processed with Xcalibur software.
Figure 6. Initial rates of reduction of 3 by 9,10-dihydroanthracene as a func-
tion of [4] (A) and [9,10-dihydroanthracene] (B) in CH3CN. Conditions: [9,10-
dihydroanthracene]=4.0ꢁ10ꢀ3 m (A); [4]=5ꢁ10ꢀ5 m (B); 0.2% H2O (v/v),
CH3CN, ꢀ408C.
fide oxidation into methyl phenyl sulfoxide which equals four
orders of magnitude.[10]
Conclusions
This study is a logical refinement of the mechanistic portrait
for the activation of CꢀH bonds of hydrocarbons by 2. It sup-
ports the conclusions of Nam and co-workers[13] concerning
homolytic CꢀH abstraction without a rebound step and further
contributes to the general understanding of hydrocarbon oxi-
dations by high-valent iron–oxo complexes. By performing
studies at lower temperatures, a tunneling effect in CꢀH bond
activation by 2 was established that contributes to higher reac-
tivity, particularly below 08C. The determination of the activa-
tion parameters, particularly of the enthalpies of activation
DHꢀ, helped to show that standard correlation of rate con-
stants with bond dissociation energies DCꢀH may be misleading
for the most general mechanistic conclusions. The DHꢀ vs.
DCꢀH correlation revealed that 9,10-dihydroanthracene reacts
differently to all other hydrocarbons studied and presumably
Hydrocarbon oxidation procedure
In a typical experiment, appropriate volumes of acetonitrile, water
(0.2% v/v), and a stock solution of 1 (1 mm) were added in
a quartz cuvette to attain the desired final concentration of 1. The
Chem. Eur. J. 2015, 21, 1803 – 1810
www.chemeurj.org
1809
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[17] M. K. Whittlesey, R. J. Mawby, R. Osman, R. N. Perutz, L. D. Field, M. P.
Wilkinson, M. W. George, J. Am. Chem. Soc. 1993, 115, 8627.
[18] K. Cho, P. Leeladee, A. J. McGown, S. DeBeer, D. P. Goldberg, J. Am.
Chem. Soc. 2012, 134, 7392–7399.
[19] J. H. Espenson, Chemical Kinetics and Reaction Mechanisms, 2nd ed.,
McGraw-Hill, Inc., New York, 1995.
[20] X. Wang, S. Peter, M. Kinne, M. Hofrichter, J. T. Groves, J. Am. Chem. Soc.
2012, 134, 12897–12900.
[21] S. J. Blanksby, G. B. Ellison, Acc. Chem. Res. 2003, 36, 255.
[22] C. Fertinger, A. Franke, R. van Eldik, J. Biol. Inorg. Chem. 2012, 17, 27.
[23] R. G. Wilkins, Kinetics and Mechanism of Reactions of Transition Metal
Complexes, 2nd ed., VCH, Weinheim, 1991.
solution was cooled at ꢀ408C for 10 min. A solution of mCPBA
(1 equiv, 2ꢁ10ꢀ2 m in acetonitrile) was added to the cold reaction
mixture to generate 2. Completion was determined by the absorb-
ance stabilization at 630 nm (lmax for 2). The final concentration of
2 was assayed from the previously determined extinction coeffi-
cients of 2 (4200mꢀ1 cmꢀ1 at 630 nm).[7] An appropriate volume of
the stock solution of the hydrocarbon in acetonitrile (0.02–1.00 m)
was then added to the pre-formed 2. Initial reaction rates were de-
termined from the absorbance decay at 630 nm (lmax for 2; see the
Supporting Information). Products produced from ethylbenzene
were quantified by quenching the reaction mixture with p-
OMeC6H4SCH3 (5ꢁ10ꢀ2 m), which is approximately 5 orders of mag-
nitude more reactive toward 2 than towards ethylbenzene.[10] The
products of oxidation of other hydrocarbons were analyzed
similarly. The reaction completion time was inferred by observing
the stabilization of the band at 630 nm.
[24] R. P. Bell, Chem. Soc. Rev. 1974, 3, 513.
[25] H. Kwart, Acc. Chem. Res. 1982, 15, 401.
[26] Cytochrome P450: Structure, Mechanism, and Biochemistry, 3rd ed. (Ed.:
P. R. Ortiz de Montelano), Kluwer Academic/Plenum Publishers, New
York, 2005.
[27] J. R. Bryant, J. M. Mayer, J. Am. Chem. Soc. 2003, 125, 10351.
[28] I. W. C. E. Arends, P. Mulder, K. B. Clark, D. D. M. Wayner, J. Phys. Chem.
1995, 99, 8182.
Keywords: CꢀH activation · iron · oxidation · oxo species ·
[29] M. J. Zdilla, J. L. Dexheimer, M. M. Abu-Omar, J. Am. Chem. Soc. 2007,
129, 11505.
reaction mechanisms
[30] T. H. Parsell, M.-Y. Yang, A. S. Borovik, J. Am. Chem. Soc. 2009, 131, 2762.
[31] E. J. Klinker, S. Shaik, H. Hirao, L. Que Jr., Angew. Chem. 2009, 121, 1317;
Angew. Chem. Int. Ed. 2009, 48, 1291.
[32] A. Streitweiser, Jr., C. M. Berke, K. Robbers, J. Am. Chem. Soc. 1978, 100,
8271.
[33] E. Buncel, B. Maenon, J. Chem. Soc. Chem. Commun. 1976, 648.
[34] A. Streitwieser, Jr., W. R. Young, R. A. Caldwell, J. Am. Chem. Soc. 1969,
91, 527.
[1] T. J. Collins, Acc. Chem. Res. 1994, 27, 279.
[2] T. J. Collins, Acc. Chem. Res. 2002, 35, 782.
[3] A. D. Ryabov, T. J. Collins, Adv. Inorg. Chem. 2009, 61, 471.
[4] A. D. Ryabov, Adv. Inorg. Chem. 2013, 65, 117.
[5] S. Kundu, M. Annavajhala, I. V. Kurnikov, A. D. Ryabov, T. J. Collins, Chem.
Eur. J. 2012, 18, 10244.
[6] A. Chanda, X. Shan, M. Chakrabarti, W. C. Ellis, D. L. Popescu, F. Tiago
de Oliveira, D. Wang, L. Que, Jr., T. J. Collins, E. Mꢂnck, E. L. Bominaar,
Inorg. Chem. 2008, 47, 3669.
[7] F. Tiago de Oliveira, A. Chanda, D. Banerjee, X. Shan, S. Mondal, L.
Que Jr., E. L. Bominaar, E. Mꢂnck, T. J. Collins, Science 2007, 315, 835.
[8] A. Ghosh, D. A. Mitchell, A. Chanda, A. D. Ryabov, D. L. Popescu, E. C.
Upham, G. J. Collins, T. J. Collins, J. Am. Chem. Soc. 2008, 130, 15116.
[9] T. J. Collins, C. Walter, Sci. Am. 2006, 294, 82.
[35] A. S. Larsen, K. Wang, M. A. Lockwood, G. L. Rice, T.-J. Won, S. Lovell, M.
ˇ
Sadꢃlek, F. Turecek, J. M. Mayer, J. Am. Chem. Soc. 2002, 124, 10112.
[36] I. C. Lewis, T. Edstrom, J. Org. Chem. 1963, 28, 2050.
[37] W. A. Duncan, J. P. Sheridan, F. L. Swinton, Trans. Faraday Soc. 1966, 62,
1090.
[38] J. H. Wengrovius, J. Sancho, R. R. Schrock, J. Am. Chem. Soc. 1981, 103,
3932.
[39] D. Usharani, D. C. Lacy, A. S. Borovik, S. Shaik, J. Am. Chem. Soc. 2013,
[10] S. Kundu, J. V. K. Thompson, A. D. Ryabov, T. J. Collins, J. Am. Chem. Soc.
2011, 133, 18546.
135, 17090.
[40] C. P. Horwitz, A. Ghosh, U. S. Patent 7,060,818, Carnegie Mellon Universi-
[11] S. Kundu, Ph. D. Thesis, Carnegie Mellon University (Pittsburgh), 2012.
[12] M. Ghosh, K. K. Singh, C. Panda, A. Weitz, M. P. Hendrich, T. J. Collins,
B. B. Dhar, S. S. Gupta, J. Am. Chem. Soc. 2014, 136, 9524.
[13] E. Kwon, K.-B. Cho, S. Hong, W. Nam, Chem. Commun. 2014, 50, 5572.
[14] G. Xue, C. Geng, S. Ye, A. T. Fiedler, F. Neese, L. Que, Jr., Inorg. Chem.
2013, 52, 3976.
ty, 2006.
[41] C. R. Goldsmith, R. T. Jonas, T. D. P. Stack, J. Am. Chem. Soc. 2002, 124,
83.
[42] W. L. F. Armarego, C. L. L. Chai, Purification of Laboratory Chemicals,
5th ed., Elsevier Science USA, 2003.
[15] G. Xue, R. D. Hont, E. Mꢂnck, L. Que Jr., Nature Chem. 2010, 2, 400.
[16] D. Wang, M. Zhang, P. Bꢂhlmann, L. Que Jr., J. Am. Chem. Soc. 2010,
132, 7638.
Received: August 27, 2014
Published online on November 19, 2014
Chem. Eur. J. 2015, 21, 1803 – 1810
www.chemeurj.org
1810
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim