Is [FeO]2+ the active center also in iron containing zeolites? A density functional theory study of methane hydroxylation catalysis by Fe-ZSM-5 zeolite

Article abstract of DOI:10.1021/ic1000073

Arguments are put forward that the active ∝-oxygen site in the Fe-ZSM-5 catalyst consists of the FeO²⁺ moiety. It is demonstrated that this zeolite site for FeO²⁺ indeed obeys the design principles for high reactivity of the FeO²⁺ moiety proposed earlier: a ligand environment consisting of weak equatorial donors (rather oxygen based than nitrogen based) and very weak or absent trans axial donor. The ∝-oxygen site would then owe its high reactivity to the same electronic structure features that lends FeO²⁺ its high activity in biological systems, as well as in the classical Fenton chemistry.

Full text of DOI:10.1021/ic1000073

3866 Inorg. Chem. 2010, 49, 3866–3880
DOI: 10.1021/ic1000073
Is [FeO]2þ the Active Center Also in Iron Containing Zeolites? A Density Functional
Theory Study of Methane Hydroxylation Catalysis by Fe-ZSM-5 Zeolite
Angela Rosa,*,† Giampaolo Ricciardi,*,† and Evert Jan Baerends*,‡,§
†
‡
ꢀ
Dipartimento di Chimica, Universita della Basilica, Via N. Sauro 85, 85100 Potenza, Italy, Dep. of Chemistry,
§
Pohang Univ. of Science and Technology, Pohang 790-784, South-Korea, and Theoretische Chemie,
Vrije Universiteit Amsterdam, De Boelelaan 1083, 1081 HV Amsterdam, The Netherlands
Received January 2, 2010
Arguments are put forward that the active R-oxygen site in the Fe-ZSM-5 catalyst consists of the FeO2þ moiety. It is
demonstrated that this zeolite site for FeO2þ indeed obeys the design principles for high reactivity of the FeO2þ moiety
proposed earlier: a ligand environment consisting of weak equatorial donors (rather oxygen based than nitrogen based)
and very weak or absent trans axial donor. The R-oxygen site would then owe its high reactivity to the same electronic
structure features that lends FeO2þ its high activity in biological systems, as well as in the classical Fenton chemistry.
Introduction
calculations9,10 and inCar-ParrinelloMD simulationsofthe
reaction of H2O2 with Fe2þ in water solvent,11,12 but the most
recent experimental work8 favors OH3 radical rather than
FeO2þ as reaction intermediate (see ref 13 for an alternative
experimental point of view). The reactivity of FeO2þ will be
determined by the ligand environment. It has been empha-
sized that the fact that FeO2þ may have different spin states is
crucial for a complete understanding of the reactivity in the
various cases, the high-spin quintet (S = 2) being notably
more active than the triplet (S=1) state. The quintet may not
be the ground state initially in, for example, the reactants
complex, but crossover to the quintet state may then offer
passage over a lower barrier, compare the two-state reactivity
documented in many cases by Shaik and co-workers.14,15 It is
however important to note that the triplet to quintet conver-
sion is only one of the electronic structure factors that
determine the FeO2þ reactivity. In Figure 1 we give the
orbital energy level diagrams16 of two prototype systems,
the pentaamine complex, representative of systems such as
[FeIV(O)(N4Py)]2þ and [FeIV(O)(TMC)(CH3CN)]2þ, and
the pentaaqua complex, present in the Fenton chemistry
and representative of the oxygen donor ligand environment
in some of the biological systems like MMO and TauD. The
diagrams contain the iron-oxo bonding levels of σ and π type
(1π and 2σ) as lowest levels (ca. 50% oxygen character, but
Ironoxo compounds play a very important role in the
catalysis of many oxidation reactions, ranging from the clas-
sical Fenton reaction to biological systems like the porphyrin
based P450 enzymes,1,2 taurine/R-ketoglutarate dioxygenase
(TauD),3,4 and methane monooxygenase (MMO).5,6 In
many cases the FeO2þ unit is the postulated active moiety.
This is the case in P450s, TauD, the synthesized biomimetic
analogues7 like [FeIV(O)(N4Py)]2þ and [FeIV(O)(TMC)-
(CH3CN)]2þ (N4Py = N,N-bis(2- pyridylmethyl)-N-bis(2-
pyridyl)methylamine, TMC=1,4,8,11-tetramethyl-1,4,8,11-
tetraazacyclotetradecane), and the recently synthesized
pentaaqua complex of FeO2þ.8 Also in the Fenton chemistry
(H2O)5FeO2þ has been found to be the active interme-
diate in electronic structure density functional theory (DFT)
*To whom correspondence should be addressed. E-mail: angela.rosa@
unibas.it (A.R.), rg010sci@unibas.it (G.R.), ej.baerends@few.vu.nl (E.J.B.).
(1) Meunier, B.; de Visser, S. P.; Shaik, S. Chem. Rev. 2004, 104, 3947.
€
(2) Schoneboom, J. C.; Cohen, S.; Lin, H.; Shaik, S.; Thiel, W. J. Am.
Chem. Soc. 2004, 126, 4017.
(3) Price, J. C.; Barr, E. W.; Tirupati, B., Jr.; J. M., B.; Krebs, C.
Biochemistry 2003, 42, 7497.
(4) Hoffart, L. M.; Barr, E. W.; Guyer, R. B.; Bollinger, J. M., Jr.; Krebs,
C. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 14738.
(5) Solomon, E. I.; Sundaram, U. M.; Machonkin, T. E. Chem. Rev. 1996,
96, 2563.
(6) Que, L.; Dong, Y. H. Acc. Chem. Res. 1996, 29, 190.
(7) Rohde, J.-U.; In, J.-H.; Lim, M. H.; Brennessel, W. W.; Bukowski,
€
M. R.; Stubna, A.; Munck, E.; Nam, W.; Que, L., Jr. Science 2003, 299, 1037.
(12) Ensing, B.; Buda, F.; Gribnau, M. C. M.; Baerends, E. J. J. Am.
Chem. Soc. 2004, 126, 4355.
(8) Pestovsky, O.; Stoian, S.; Bominaar, E. L.; Shan, X.; Munck, E.; Que,
L.; Bakac, A. Angew. Chem., Int. Ed. 2005, 44, 6871.
(9) Buda, F.; Ensing, B.; Gribnau, M. C. M.; Baerends, E. J. Chem.;Eur.
J. 2001, 7, 2775.
(10) Buda, F.; Ensing, B.; Gribnau, M. C. M.; Baerends, E. J. Chem.;
Eur. J. 2003, 9, 3436.
(11) Ensing, B.; Buda, F.; Blochl, P. E.; Baerends, E. J. Phys. Chem.
(13) Kremer, M. Int. J. Chem. Kin. 2006, 38, 725.
€
(14) Shaik, S.; Filatov, M.; Schroder, D.; Schwarz, H. Chem.;Eur. J.
1998, 4, 193.
(15) Hirao, H.; Kumar, D.; Que, L.; Shaik, S. J. Am. Chem. Soc. 2006,
128, 8590.
€
(16) Bernasconi, L.; Louwerse, M. J.; Baerends, E. J. Eur. J. Inorg. Chem.
2007, 3023.
Chem. Phys. 2002, 4, 3619.
r
pubs.acs.org/IC
Published on Web 03/21/2010
2010 American Chemical Society

Article
Inorganic Chemistry, Vol. 49, No. 8, 2010 3867
Figure 1. Orbital energy diagram showing the strong stabilization of the levels of the all oxygen donor ligand environment of (H2O)5FeO2þ(S = 2)
compared to the all nitrogen donor ligand environment of (NH3)5FeO2þ (S=1). Unpaired electrons are indicated in red. Spin R orbital levels are stabilized
by the exchange field compared to the corresponding β spin orbitals.
nominally the oxo levels), and above those the set of nomi-
nally 3d levels. We wish to draw attention to the enormous
difference in the R-3σ* (R-3dz2) orbital energy in the two
diagrams (ca. 3 eV), making the R-3σ* of the (H2O)5FeO2þ
by far the stronger electron acceptor.
the N donor ligands) depends on the ligand environment.10,17-19
We note that the balance between the R-σ* FMO channel
and the β-π* channel19 for oxidation reactions also depends
on the metal used, as has been demonstrated in the series of
complexes (H2O)5MO2þ (M = V-Ni).20 It is obvious that
the energy of the R-3σ* can be further lowered if the trans
axial ligand is very weak or absent. Therefore the design
principles for a highly active FeO2þ hydroxylation catalyst
are as follows:16 (a) equatorial ligands should be weak
donors, for example, preferably not the N lone pair donors
used widely in the biomimetic analogues, but rather the
weaker oxygen lone pair donors; (b) a weak or preferably
absent trans axial donor. Very recently Que and co-workers
have synthesized an interesting modification of the generally
used nitrogen donor based pentacoordinate ligands like
TMC, which have an equatorial square of four N donors
and a trans axial N donor.21 By replacing it with TMG3tren
(a tetradentate ligand that coordinates the iron with three
equatorial and one trans axial N lone pair), the difference in
the destabilizations of the equatorial dx2-y2 and dxy is lifted,
these orbitals becoming degenerate by symmetry. In this
ligand environment it is easier for the FeO2þ unit to retain
the high-spin configuration, which indeed it does. On the
other hand, the ligand environment still consists of all strong
N donors, including a strong trans axial donor, destabilizing
the R-dz2. Indeed, the reactivity of this newly synthesized
complex was not extreme. We note that biological systems
Obviously, the replacement of N donor ligands by O donor
ligands causes a large stabilization of the R-3σ*. The triplet to
quintet spin state change contributes to this effect, but is not
solely responsible. Emptying the R-dx2-y2, when it is being
pushed up too high by strong equatorial ligands like NH3
groups, and occupying the β-dxy, reduces the number of
unpaired spins from 4 in the quintet (H2O)5FeO2þ to only
2 in the triplet (NH3)5FeO2þ. The stabilizing exchange field
for the R spin levels then becomes much weaker. When the
R-dx2-y2 electron isflipped to β and putinto β-dxy, that is, spin
paired with R-dxy to form the (dxy)2 pair, without changing
the ligands, the net loss of stabilization of R-3σ* is about 1 eV,
one-third of the totalshift of about 3 eVdisplayed inFigure 1.
The remainder of the shift comes from the stronger donation
by the N lone pairs, both in terms of interaction strength
(governed by Hamiltonian matrix element and fragment
orbital energy difference) and in terms of charge effect.
It is clear that the relative importance of the R-3σ* as accep-
tor orbital (very important in the Fenton complex (H2O)5-
FeO2þ compared to the β-2π* orbitals (more important with
(17) Neidig, M. N.; Decker, A.; Choroba, O. W.; Huang, F.; Kavana, M.;
Moran, G. R.; Spencer, J. B.; Solomon, E. I. Proc. Natl. Acad. Sci. U.S.A.
2006, 103, 12966.
(18) Louwerse, M. J.; Baerends, E. J. Phys. Chem. Chem. Phys. 2007, 9,
156.
(19) Decker, A.; Rohde, J.-U.; Klinker, E. J.; Wong, S. D.; Que, L.;
Solomon, E. I. J. Am. Chem. Soc. 2007, 129, 15983.
(20) Michel, C.; Baerends, E. J. Inorg. Chem. 2009, 48, 3628.
(21) England, J.; Martino, M.; Farquhar, E. R.; Frisch, J. R.; Bominaar,
€
E. L.; Munck, E.; Que, L., Jr. Angew. Chem., Int. Ed. 2009, 48, 3622.

3868 Inorganic Chemistry, Vol. 49, No. 8, 2010
Rosa et al.
like TauD and MMO are characterized by more oxygen than
nitrogen donor ligands.
Fe complexes show very high reactivity and produce the
same products via the same type of mechanism. For this
reason such materials were called “mineral enzymes” or
“zeozymes”.37,38
On the basis of the design principles put forward above,
EDTA has been proposed as a ligand that should be very
effective, combining equatorial oxygen donors with a weak
trans axial donor (two off-axis N lone pairs at large dis-
tance).16 In the present contribution we will propose that the
catalytic activity of the well-known R-oxygen in Fe-ZSM-5
can be considered from the same point of view: it possibly
owes its high oxidative power to the presence of a FeO2þ unit
in exactly the favorable ligand environment of equatorial
oxygen ligands and absence of a trans axial donor.
The location and nuclearity of R-sites (an R-site is defined
as an iron entity able to adsorb one atom of R-oxygen), have
been extensively debated.38 It is by now widely accepted that
the R-sites are located inside the channels of the zeolite
matrix.39-42 The nuclearity of the R-sites is less clear,
although significant evidence has been given that they are
monatomic and in a paired arrangement, with each of the Fe
ions in the binuclear structure being capable of generating
R-oxygen species independently.23 Although the oxygen
species involved in the C-H bond activation have not been
identified experimentally, [FeO]þ, [Fe(O)2]þ, [OFeO]þ, and
[FeO]2þ species have been proposed as the catalytically active
centers on the basis of quantum chemical calculations.43-48
Depending on the ironoxo species assumed as the catalytic
active center, different C-H bond activation mechanisms
have been theoretically proposed.
Iron exchanged ZSM-5 has been shown to be an active
catalyst for many reactions, a very interesting example being
the decomposition of N2O. This process has been suggested
to proceed via oxidation of Fe2þ to Fe3þ, forming active
oxygen species called by Panov and co-workers atomic R-
oxygen.22,23 This R-oxygen species is capable of selectively
and quantitatively converting benzene to phenol at ambient
temperature,24,25 similar to the active oxygen of monooxy-
genases (MMOs), forwhich the hydroxylation of aromaticsis
a typical reaction.26 The R-oxygens in Fe-ZSM-5 zeolites
react at room temperature even with methane, the most inert
organic molecule, giving quantitative yield of methanol,27
thus challenging the reactivity of the active oxygen of
methane monooxygenases (MMO).28-33 Dubkov et al.34
studied the kinetic isotope effect (KIE = kH/kD) for the
methane and benzene oxidations by R-oxygen and found
high KIE values (1.9-5.5, depending on the reaction tem-
perature) for methane oxidation and no isotopic effect for
benzene oxidation. This means that, similar to the MMO
case, methane oxidation by R-oxygen involves the cleavage of
the C-H bond as the rate-limiting step. The oxidation of
benzene proceeds somewhat differently. Understandably,
benzene is attacked by the electrophilic FeO2þ at its π elec-
tron cloud, as has been found for benzene hydroxylation with
biomimetic non-heme FeO2þ complexes35 and with a zeolite
[FeO]þ with a d5 Fe(III) has been assumed as the active
catalytic center by Yoshizawa et al.43 We wish to stress that
the [FeO]þ would have to accommodate the extra electron it
possesses compared to [FeO]2þ in the R-dz2 orbital, which we
view16,18 as the prime acceptor orbital of this moiety, respon-
sible for its extreme electrophilicity and high activity in C-H
functionalization. This will lead to much higher barriers than
expected for [FeO]2þ. The authors of ref 43 studied the
reaction pathways and the energetics for the direct hydro-
xylation of methane and of benzene over [FeO]þ in the Fe-
ZSM-5 zeolite surface. The barrier height computed for the
second transition state in the two-step process, TS2, is quite
high (41.6 and 31.1 kcal/mol in the case of methane and
benzene, respectively). The proposed mechanism does not
account for the high KIE values measured for methane
oxidation by zeolite R-oxygen.34 A d5 Fe(III) iron center
with occupied R-dz2 is also advocated in the theoretical work
by Malykhin et al.,48 who propose an electronic configura-
cluster with FeO2þ 36
. Therefore, in many respects, R-oxygen
in Fe-ZSM-5 zeolites exhibits an exciting analogy to the active
oxygen in MMO: in both cases oxygen species coordinated to
tion [Fe(III)O ]
- 2þ. These authors note however that in their
3
calculations on the model system FeO(OH)2, the lowest
energy configuration at the minimum energy geometry does
have a d4 Fe(IV) iron center, that is, empty R-dz2. Important
recent experimental evidence, based on resonant inelastic
(22) Panov, G. I.; Uriarte, A. K.; Rodkin, M. A.; Sobolev, V. I. Catal.
Today 1998, 41, 365.
(23) Dubkov, K. A.; Ovanesyan, N. S.; Shteinman, A. A.; Starokon,
E. V.; Panov, G. I. J. Catal. 2002, 207, 341.
(24) Sobolev, V. I.; Kharitonov, A. S.; Paukshtis, Y. A.; Panov, G. I.
J. Mol. Catal. 1993, 84, 117.
(25) Panov, G. I.; Sobolev, V. I.; Dubkov, K. A.; Parmon, V. N.;
Ovanesyan, N. S.; Shilov, A. E.; Shteinman, A. A. React. Kinet. Catal. Lett.
1997, 61, 251.
(26) Shilov, A. E. Metal Complexes in Biomimetic Chemical Reactions;
CRC Press: New York, 1997.
(27) Panov, G. I.; Sobolev, V. I.; Dubkov, K. A.; Kharitonov, A. S. Stud.
Surf. Sci. Catal. 1996, 101, 493.
(28) Dewitt, J. G.; Bentsen, J. G.; Rosenzweig, A. C.; Hedman, B.; Green,
J.; Pilkington, S.; Papaefthymion, G. C.; Dalton, H.; Hodgson, K. O.;
Lippard, S. J. J. Am. Chem. Soc. 1991, 113, 9219.
(29) Shteinman, A. A. Russ. Chem. Bull. 2001, 50, 1795.
(30) Que, L. Pure Appl. Chem. 1998, 70, 947.
(31) Kopp, D. A.; Lippard, S. J. Curr. Opin. Chem. Biol. 2002, 6, 568.
(32) Siegbahn, P. E. M.; Crabtree, R. H.; Nordlund, P. J. Biol. Inorg.
Chem. 1998, 3, 314.
(37) Parton, R.; De Vos, D.; Jakobs, P. A. In Zeolite Micropores Solids:
Synthesis, Structure and Reactivity; Derouane, E. G., Ed.; Kluwer: The Nether-
lands, 1992; p 555.
(38) Panov, G. I. CATTECH 2000, 4, 18.
(39) Ribera, A.; Arends, I. W. C. E.; de Vries, S.; Perez-Ramires, J.;
Sheldon, R. A. J. Catal. 2000, 195, 287.
(40) Pirutko, L. W.; Parenago, O. O.; Lunina, E. V.; Kharitonov, A. S.;
Okkel, L. G.; Panov, G. I. React. Kinet. Catal. Lett. 1994, 52, 275.
(41) Panov, G. I.; Kharitonov, A. S.; Fenelonov, V. B.; Voskresenskaya,
T. P.; Rudina, N. A.; Molchanov, V. V.; Plyasova, M. L. Zeolites 1995, 15,
253.
(42) Pirutko, L. W.; Dubkov, K. A.; Solovjeva, L. P.; Panov, G. I. React.
Kinet. Catal. Lett. 1996, 58, 105.
(43) Yoshizawa, A. L.; Shiota, Y.; Yumura, T.; Yamabe, T. J. Phys.
Chem. B 2000, 104, 734.
(44) Kachurovskaya, N. A.; Zhidomirov, G. M.; Hensen, E. J. M.; van
Santen, R. A. Catal. Lett. 2003, 86, 25.
(33) Lieberman, R. L.; Rosenzweig, A. C. Nature 2005, 34, 177.
(34) Dubkov, K. A.; Sobolev, V. I.; Talsi, E. P.; Rodkin, M. A.; Watkins,
N. H.; Shteinman, A. A.; Panov, G. I. J. Mol. Catal. 1997, 123, 155.
(35) de Visser, S. P.; Oh, K.; Han, A.-R.; Nam, W. Inorg. Chem. 2007, 46,
4632.
(36) Fellah, M. F.; van Santen, R. A.; Onal, I. J. Phys. Chem. C 2009, 113,
15307.
(45) Kachurovskaya, N. A.; Zhidomirov, G. M.; van Santen, R. A.
J. Phys. Chem. B 2004, 108, 5944.
(46) Ryder, J. A.; Chakraborty, A. K.; Bell, A. T. J. Catal. 2003, 220, 84.
(47) Liang, W.-Z.; Bell, A. T.; Head-Gordon, M.; Chakraborty, A. K.
J. Phys. Chem. B 2004, 108, 4362.
(48) Malykhin, S.; Zilberberg, I.; Zhidomirov, G. M. Chem. Phys. Lett.
2005, 414, 434.

Article
Inorganic Chemistry, Vol. 49, No. 8, 2010 3869
X-ray scattering, has been put forward that supports d5
Fe(III) rather than d4 Fe(IV) in the zeolite.49,50 In theoretical
work on the decomposition of N2O over Fe exchanged zeolite
Bell et al. have assumed a FeOþ active center to start with,
which they showed may lead to the peroxo species [FeO2]þ
and the dioxo species [OFeO]þ as active sites for methane
hydroxylation.47
functional provides quite accurate geometries and bond ener-
gies of first-row transition metal complexes,63-65 but fails in
describing close-lying spin states, in particular in iron com-
plexes.64,66,67 On the other hand, the OPBE functional has
been shown to yield correct spin state relative stabilities for
various inorganic and organometallic iron complexes,68-71
but fails in describing weak bonds, including hydrogen bonds
and π-π stacking interactions.64,66,67,72 For sake of compar-
ison with previousstudies ontheC-H activation reactivity of
Fe-ZSM-5 zeolites, the calculations on the catalytic active
iron(IV)-oxo species in the high-spin (S = 2) and low-spin
(S=1) states were also performed at the B3LYP59,73,74 level
of theory, using TURBOMOLE. All calculations were per-
formedinthegas-phaseusing thespin-unrestrictedapproach.
The BP86 and OPBE ADF calculations included relativistic
effects through the Zero-Order Regular Approximation
(ZORA) and made use of a basis set of Slater type orbitals
of TZ2P quality for all atomic species. The BP86 and B3LYP
TURBOMOLE calculations were performed with the recent
triple-ζ all-polarizable basis set, def2-TZVP,75 which pro-
vides a balanced description of all elements of the periodic
table. In the BP86 TURBOMOLE calculations, the resolu-
tion of the identity (density fitting) approach was used to save
computer time.76
Transitionstateswerefoundbyfirstperforming apotential
energy scan along the pertinent reaction coordinate. The
maximum energy structures were then taken as a guess for
the transition state search. Frequency calculations were per-
formed to characterize local minima and transition states.
The transition states described here had one imaginary fre-
quency for the correct mode, whereas all local minima had
real frequencies only. Zero-point energies (ZPE) corrections
were obtained from the frequency calculations and are inclu-
ded in the reaction energies and activation energies through-
out.
[FeO]2þ has long been known to be the active center in
heme enzymes such as cytochrome P4501,2 and has been
suggested to be also responsible for the catalytic activity of
Fe-ZSM-5 zeolites.22,51 [FeO]2þ with a d4 Fe(IV) iron
center has been considered as the active catalytic species by
Kachurovskaya et al.44,45 and Fellah et al.36 in their DFT
studies of the benzene-to-phenol oxidation reaction on iron
exchanged zeolites. These authors proposed a reaction path
for the benzene oxidation that is consistent with the experi-
mental evidence for the absence of the kinetic H/D isotope
effect in phenol formation.
In this work we study the methane hydroxylation reaction
catalyzed by the ferryl ion [FeO]2þ in the active center on Fe-
ZSM-5 zeolites. Specifically, we want to investigate whether
the ligand environments offered by plausible cation exchange
sites of the ZSM-5 zeolite conform to the requirements put
forward above for high activity of [FeO]2þ as a hydroxylation
catalyst. We will come to the conclusion that the design
principles for optimal activity of [FeO]2þ, which are weak
equatorial donor ligands, and a very weak or absent trans
axial donor, are perfectly met by the ligand environments for
[FeO]2þ in this case. This connects the Fe-ZSM-5 zeolite
catalyst for methane hydroxylation immediately with the
active biological systems for this reaction, such as TauD and
MMO. The rebound mechanism proposed by Groves52,53
and generally accepted for C-H oxidation by metal oxo
complexes also applies in this case.
The kinetic isotope effect (KIE) of replacing hydrogen by
deuterium atoms in the methane was estimated using the
semiclassical model based on Eyring’s eq 1.
Methods
kH=kD ¼ exp½-ðΔGqH - ΔGDq Þ=RTꢀ
ð1Þ
The calculations were done with ADF (Amsterdam Den-
sity Functional)54-56 as well as with TURBOMOLE V6.057,58
packages. Two GGA (Generalized Gradient Approximation)
exchange-correlation functionals were employed, BP8659,60
and OPBE.61,62 The use of both these functionals is needed to
provide reliable prediction of both the geometries and spin-
state energetics of the reaction intermediates. The BP86
As the KIEs originate, in this framework, from isotopic ZPE
differences between the reactant (R) and transition state, the
thermal factors playing a very minor role, the (ΔGqH-ΔGDq )
term was approximated by the (ZPEqH - ZPEHR - ZPEDq
þ
ZPERD) term.
Model System. In our model of the “R-oxygen” species, ZFeO
(Z = zeolite), the zeolite is represented by the dianionic clus-
ter previously used by Kachurovskaya et al.44 This cluster is
(49) Pirngruber, G. D.; Grunwaldt, J. D.; van Bokhoven, J. A.; Kalytta,
A.; Reller, A.; Safonova, O. V.; Glatzel, P. J. Phys. Chem. B 2006, 110, 18104.
(50) Pirngruber, G. D.; Grunwaldt, J. D.; Roy, P. K.; van Bokhoven,
J. A.; Safonova, O. V.; Glatzel, P. Catal. Today 2007, 126, 127.
(51) Kiwi-Minsker, L.; Bulushev, D. A.; Renken, A. J. Catal. 2003, 219,
273.
(52) Groves, J. T.; McClusky, G. A. J. Am. Chem. Soc. 1976, 98, 859.
(53) Groves, J. T. J. Chem. Educ. 1985, 62, 928.
(54) Fonseca Guerra, C.; Snijders, J. G.; te Velde, G.; Baerends, E. J.
Theor. Chem. Acc. 1998, 99, 391.
(55) ADF2008.01; SCM, Theoretical Chemistry, Vrije Universiteit Amster-
dam: The Netherlands; http://www.scm.com.
(56) te Velde, G.; Bickelhaupt, F. M.; Baerends, E. J.; Fonseca Guerra,
C.; van Gisbergen, S. J. A.; Snijders, J. G.; Ziegler, T. J. Comput. Chem. 2001,
22, 931.
(63) Jensen, K. P.; Roos, B. O.; Ryde, U. J. Chem. Phys. 2007, 126,
014103.
(64) Neese, F. J. Biol. Inorg. Chem. 2006, 11, 702.
(65) Berry, J. F.; DeBeer, G. S.; Neese, F. Phys. Chem. Chem. Phys. 2008,
10, 4361.
(66) Ghosh, A. J. Biol. Inorg. Chem. 2006, 11, 712.
(67) Radon, M.; Pierloot, K. J. Phys. Chem. A 2008, 112, 11824.
ꢁ
(68) Fouqueau, A.; Mer, S.; Casida, M. E.; Daku, L. M. L.; Hauser, A.;
Mineva, T.; Neese, F. J. Chem. Phys. 2004, 120, 9473.
(69) Fouqueau, A.; Casida, M. E.; Lawson, L. M.; Hauser, A.; Neese, F.
J. Chem. Phys. 2005, 122, 044110.
(70) Swart, M.; Groenhof, A. R.; Ehlers, A. W.; Lammertsma, K. J. Phys.
Chem. A 2004, 108, 5479.
(71) Conradie, J.; Quarless, D.; Hsu, H.-F.; Harrop, T.; Lippard, S. J.;
Koch, S.; Ghosh, A. J. Am. Chem. Soc. 2007, 129, 10446.
ꢀ
(72) Swart, M.; Sola, M.; Bickelhaupt, F. M. J. Chem. Phys. 2009, 131,
(57) TURBOMOLE V6.0 2009; a development of University of Karlsruhe
and Forschungzentrum Karlsruhe GmbH, 1989-2007; TURBOMOLE GmbH:
Karlsruhe, Germany, 2007; available from http://www.turbomole.com.
€
€
€
(58) Ahlrichs, R.; Bar, M.; Haser, M.; Horn, H.; Kolmel, C. Chem. Phys.
Lett. 1989, 162, 165.
094103.
(59) Becke, A. Phys. Rev. A 1988, 38, 3098.
(60) Perdew, J. P. Phys. Rev. B 1986, 33, 8822.
(61) Cohen, A. J.; Handy, N. C. Mol. Phys. 2001, 99, 607.
(62) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1997, 78,
1396.
(73) Becke, A. D. J. Chem. Phys. 1993, 98, 5648.
(74) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785.
(75) Weigend, F.; Ahlrichs, R. Phys. Chem. Chem. Phys. 2005, 7, 3297.
(76) Eichkorn, K.; Treutler, O.; Haser, M.; Ahlrichs, R. Chem. Phys. Lett.
€
1995, 240, 283.

3870 Inorganic Chemistry, Vol. 49, No. 8, 2010
Rosa et al.
Figure 2. Zeolite cluster model. Color code: aluminum (green), silicon
(sand), oxygen (red), hydrogen (white).
Figure 4. BP86/ZORA/TZ2P optimized structures of the 1 and 2 iso-
mers of ZFeO in the triplet spin state. Color code as in Figure 3.
cavity of the zeolite six-membered ring, as shown in Figures 2
and 3. The structure of the ZFeO cluster as well as of the species
involved in the methane hydroxylation reaction were optimized
without constraints except for the positions of the boundary
hydrogen atoms at the -OSiH3 groups, which were kept fixed.
For the ZFeO cluster the S = 0, 1, 2, and 3 spin states were
considered. The S = 2 spin state was found to be the ground
state, with the S=1 being the next higher spin state (vide infra).
The S = 0 and S = 3 spin states were found much higher in
energy than the quintet state; therefore we did not pursue these
spin surfaces further, and the complete reaction profile for
methane hydroxylation was studied only on the triplet and
quintet spin state surfaces.
Results and Discussion
Molecular and Electronic Structure of the Fe-ZSM-5
“r-oxygen” Model. Two coordination geometries of the
iron atom to the oxygen atoms of the zeolite ring were
considered, 1 and 2. In 1 the iron coordinates to the four
oxygen atoms bonded to the Al atoms, as in the ZFeO
cluster model by Kachurovskaya et al.44 In 2 the iron
atom coordinates to three oxygen atoms bonded to the Al
atoms and an oxygen atom bonded only to Si atoms. This
type of coordination has been reported for the zinc ion in
a six-membered zeolite ring similar to that of our cluster
model.78
The S=2 and S=1 optimized structures of 1 and 2 are
displayed in Figures 3 and 4, respectively.
The important geometrical parameters, relative ener-
gies, Mulliken spin and charge populations computed at
different levels of theory are reported in Tables 1 and 2.
Figure 3. BP86/ZORA/TZ2P optimized structures of the 1 and 2 iso-
mers of ZFeO in the quintet spin state. Color code: iron (rust brown),
aluminum, silicon, oxygen, and hydrogen as in Figure 2.
constructed from two intersecting five-membered zeolite rings.
Two distant Si atoms were replaced by Al atoms, so that in the
surface a six-membered ring of two Al’s and four O’s is formed
(Figure 2). The initial coordinates of the cluster were taken from
crystallographic data for the R-position of MFI.77 In the six-
membered ring, hydrogen atoms were used to saturate broken
Si-O bonds, while Al atoms were saturated by -OSiH3 groups.
The positions of the boundary hydrogen atoms were optimized
keeping frozen the positions of all the other atoms and with
saving the crystal structure directions of the boundary bonds.
The doubly negative charge of the two Al-substituted T-sites is
compensated by the iron oxo species, [FeO]2þ, placed in the
(78) Shubin, A. A.; Zhidomirov, G. M.; Yakovlev, A. L.; van Santen,
(77) Chao, K.; Lin, J.; Wang, Y.; Lee, G. H. Zeolites 1986, 6, 35.
R. A. J. Phys. Chem. B 2001, 105, 4928.

Article
Inorganic Chemistry, Vol. 49, No. 8, 2010 3871
Table 1. Geometrical Parameters (in A), Spin Densities (ζ), Charges (q), and Relative Energies (in kcal mol-1) of the S = 2 Structures 1 and 2 as a Model for R-oxygen in
˚
Fe-ZSM-5 Zeolites
1
2
S = 2
BP86a
BP86b
OPBEa
B3LYPb
BP86a
BP86b
OPBEa
B3LYPb
Fe-Oax
Fe-O1
Fe-O2
Fe-O3
Fe-O4
Fe-OSi
1.603
1.994
2.184
2.050
2.251
1.605
2.015
2.181
2.081
2.247
1.581
1.985
2.206
2.043
2.290
1.589
2.020
2.172
2.080
2.253
1.607
1.991
2.085
2.158
1.609
2.016
2.090
2.179
1.585
1.995
2.068
2.134
1.590
2.021
2.083
2.145
2.277
0.53
2.262
0.55
2.343
0.44
2.270
0.56
c
ΔFe
0.41
0.41
0.0
0.32
0.46
0.0
ΔEd
þ 0.5
þ 0.7
þ 0.0
þ 0.3
0.0
þ 0.01
ζFe
3.07
0.62
3.06
0.64
3.15
0.61
3.30
0.43
3.13
0.62
3.08
0.63
3.21
0.60
3.31
0.41
ζO
qFe
ax
þ 1.03
þ 0.70
þ 1.07
þ 0.90
þ 1.03
þ 0.78
þ 1.08
þ 0.97
qO
-0.36
-0.31
-0.39
-0.34
-0.38
-0.32
-0.41
-0.33
ax
a ADF (ZORA/TZ2P). b Turbomole (def2-TZVP). c Displacement of the iron atom from the plane defined by the four equatorial oxygen atoms.
d Energies include ZPE correction.
Table 2. Geometrical Parameters (in A), Spin Densities (ζ), Charges (q), and Relative Energies (in kcal mol-1) of the S = 1 Structures 1 and 2 as a Model for R-oxygen in Fe-
˚
ZSM-5 Zeolites
1
2
S = 1
BP86a
BP86b
OPBEa
B3LYPb
BP86a
BP86b
OPBEa
B3LYPb
Fe-Oax
Fe-O1
Fe-O2
Fe-O3
Fe-O4
Fe-OSi
1.603
1.965
2.040
2.030
2.107
1.605
1.972
2.046
2.045
2.114
1.578
1.951
2.025
2.014
2.130
1.584
1.973
2.053
2.034
2.134
1.574
2.058
1.929
2.006
1.574
2.100
1.943
2.020
1.550
2.071
1.892
1.958
1.564
2.094
1.941
1.981
2.405
0.50
2.376
0.53
2.487
0.43
2.332
0.50
c
ΔFe
0.30
0.0
0.29
0.0
0.23
0.0
0.33
0.0
ΔEd
þ 4.6
þ 5.1
þ 2.7
þ 6.6
ζFe
1.24
0.79
1.24
0.81
1.28
0.78
1.18
0.85
1.54
0.49
1.59
0.49
1.60
0.46
1.33
0.67
ζO
qFe
ax
þ 0.94
þ 0.52
þ 0.95
þ 0.72
þ 0.97
þ 0.69
þ 1.01
þ 0.85
qO
-0.37
-0.30
-0.39
-0.32
-0.36
-0.32
-0.38
-0.33
ax
ΔE(T-Q)d,e
þ 2.7
þ 2.5
þ 11.7
þ 12.1
þ 7.7
þ 7.3
þ 15.2
þ 18.7
a ADF (ZORA/TZ2P). b TURBOMOLE (def2-TZVP). c Displacement of the iron atom from the plane defined by the four equatorial oxygen atoms.
d Energies include ZPE correction. e Energy difference between the triplet and quintet spin states.
Table 2 also reports for each structure the energy differ-
ence between the triplet and quintet spin states, ΔE
(T-Q). Inspection of the ΔE (T-Q) values reveals that
the quintet spin state is invariably lower in energy than the
triplet state for both structures, particularly for 2.
The high-spin (quintet) state is favored by the rather
weak σ-donor ability of the O-based equatorial ligands.16,79
The energy gap to the first excited spin state, the S = 1
state, depends on the density functional used, being at the
BP86 level significantly smaller than at the OPBE and
B3LYP level.
the S=1 and S=2 states likely lies somewhere in between
the calculated values. The OPBE functional has been
found to perform relatively well for the singlet-triplet
splitting in iron compounds.70 It is worth noting that the
triplet-quintet splitting is at the OPBE level only slightly
smaller than at the B3LYP level, which is in line with
recent results by Hirao et al.15 on various oxo-iron com-
plexes.
Quintet Species 51 and 52. According to the calculated
energies, the 1 and 2 isomers are nearly degenerate in their
quintet ground spin state. Thus, 51 and 52 are both
plausible models for the catalytic site. In the first excited
spin state the energies of the 1 and 2 structures lie within
3-7 kcal/mol, with the more symmetric isomer 1 being
the preferred one, particularly at the B3LYP and BP86
level of theory. As the crystal structure directions of the
boundary bonds are preserved in our cluster model, the
structural variability observed for ZFeO is not an artifact
of the chosen model. Rather, it is indicative of the
inherent flexibility of the six-membered zeolite ring. This
flexibility is expected to be enhanced in the real system
Several studies have established that while pure func-
tionals, like BP86, overly stabilize low-spin states, the
hybrid functional B3LYP in general favors high spin
states.19,80-83 Thus, the actual energy difference between
(79) Neese, F. J. Inorg. Biochem. 2006, 100, 716.
(80) Paulsen, H.; Duelund, L.; Winkler, H.; Toflund, H.; Trautwein,
H. X. Inorg. Chem. 2001, 40, 2201.
(81) Salomon, O.; Reiher, M.; Hess, B. A. J. Chem. Phys. 2002, 117, 4729.
(82) Reiher, M. Inorg. Chem. 2002, 41, 6928.
(83) Vargas, A.; Zerara, M.; Krausz, E.; Hauser, A.; Daku, L. M. L.
J. Chem. Theory Comput. 2006, 2, 1342.

3872 Inorganic Chemistry, Vol. 49, No. 8, 2010
Rosa et al.
Figure 5. Energy level diagram and molecular orbitals of 3,51 obtained from DFT/ZORA/BP86/TZ2P calculations.
where the boundary Si-H bonds are replaced by the more
flexible Si-O bonds.
bond is formed by a σ interaction between the (nomi-
nally occupied) O-2pz orbital and the (nominally un-
occupied) Fe-dz2 orbital, and a π interaction between
the (nominally occupied) O-2px,y and the (nominally half
occupied) Fe-dxz,yz orbitals, cf. the 1σ and 1πx,y doubly
occupied orbitals in Figure 1. In the diagrams of Figures 5
and 6 only the antibonding orbitals (which are nominally
the d orbitals) of σ and π symmetry are reported.
The oxygen contribution to these orbitals ranges from
36% to 46%.
Of the Fe-dδ orbitals, the half occupied dxy is substan-
tially a nonbonding orbital, whereas the dx2-y2 with lobes
along the axes is pushed up by antibonding interaction
with the σ lone pairs of the equatorial oxygen ligands10
(but much less so than with N donor ligands16). The
interaction with the equatorial ligands is stronger in 1
than in 2, as indicated by the energies of the antibonding
dx2-y2, the R and β components of which shift to higher
energy by 0.25 and 0.30 eV, respectively. This is consistent
with two of the equatorial Fe-O bonds being in the
isomer 1 significantly shorter than in 2. The stronger
interaction of the zeolite ring with the iron center in 1, is
not reflected, however, in an increased charge donation to
the iron, as the iron charge decreases only marginally on
going from 2 to 1 (see Table 1). It is clear from the level
schemes of Figures 5 and 6 that in both isomers the lowest
acceptor molecular orbital, that is, the one that can accept
density from the substrate C-H bond is the R component
of the dz2 (σ*). The surplus of R electrons creates a
strongly stabilizing exchange potential pulling all R levels,
including the dz2 (σ*), very much down compared to the
corresponding β levels. Notably, the R-dz2 (σ*) is in 2
Focusing on the geometrical details of 1 and 2 in the
quintet ground spin state (Figure 3), both isomers show a
˚
Fe-Oax distance of about 1.6 A (see Table 1), which is
typical for iron(IV)-oxo complexes7,84 with the iron lying
˚
0.3-0.6 A above the plane of the equatorial oxygen
atoms. As can be inferred from the data in Table 1, the
spin population at the iron and oxo-ligand is scarcely
influenced by the geometrical changes in the equatorial
coordination and compares well to that computed for
other S = 2 iron(IV)-oxo complexes.85,86 In the more
symmetric isomer 1, two trans Fe-O bonds, Fe-O1 and
Fe-O3, are significantly shorter than the other two,
Fe-O2 and Fe-O4. In the less symmetric isomer 52 there
are also two short and two long Fe-O bonds, the cis
Fe-O1 and Fe-O2, and the cis Fe-O3 and Fe-OSi
bonds, respectively.
Complexes 1 and 2 in the quintet ground spin state have
qualitatively the same electronic structure and bonding
scheme. They retain the distinctive electronic structure
features of most S=2 iron(IV)-oxo complexes studied to
date, as inferred from Figures 5 and 6 where the spin
unrestricted energy level diagrams and relevant molecular
orbitals of 3,51 and 3,52 are displayed.
The Fe(IV)dO (S = 2) is a d4 system with a (dxy)1-
(dx2-y2)1(dxz)1(dyz)1 ground configuration. The Fe-O
(84) Riggs-Gerlasco, P. J.; Price, J. C.; Gruyer, R. B.; Brehm, J. H.; Bart,
E. W.; Bollinger, J. M., Jr.; Krebs, C. J. Am. Chem. Soc. 2004, 126, 8108.
(85) Sinnecker, S. S. N.; Barr, E. W.; Ye, S.; Bollinger, J. M., Jr.; Neese, F.;
Krebs, C. J. Am. Chem. Soc. 2007, 129, 6168.
(86) Decker, A.; Solomon, E. I. Angew. Chem., Int. Ed. 2005, 44, 2252.

Article
Inorganic Chemistry, Vol. 49, No. 8, 2010 3873
Figure 6. Energy level diagram and molecular orbitals of 3,52 obtained from DFT/ZORA/BP86/TZ2P calculations.
Table 3. Energy Data (in kcal/mol) for the Methane Hydroxylation Reaction by 3,51 and 5
2
ΔEq
ΔE1
BP86d/BP86e
ΔEq
ΔE2
BP86d/BP86e
ΔEc
BP86d/BP86e
a
b
TSH
TSreb
BP86d/BP86e
OPBE
OPBE
BP86d/BP86e
OPBE
OPBE
51
52
31
6.6/6.4
5.2/4.3
18.8/18.2
6.9
6.0
18.7
6.5/6.2
5.0/4.1
18.5/18.0
6.5
5.8
18.5
1.9/2.1
2.0/2.0
3.7/3.5
-41.2/-44.5
-39.9/-43.4
-34.4/-39.3
-38.7
-37.6
-35.0
-10.9/-13.8
-7.8/-12.1
-2.3/-4.8
-20.1
-18.5
-10.1
a ΔE1 = E(I) - E(R). b ΔE2 = E(PC) - E(I). c ΔE = E(P) - E(R). d ADF (ZORA/TZ2P). e TURBOMOLE (def2-TZVP).
3
about 0.2 eV lower in energy than in 1, on account of the
reduced equatorial field in the former. As visible in the
plots of the dz2 (σ*) in Figures 5 and 6, this orbital has
some antibonding character with the equatorial oxygen
lone pairs. The empty β-dxy, which is actually lower in
energy than the R-dz2 (σ*), in both 1 and 2, and the
empty β-dx2-y2 will not play a role since they are not
involved in the Fe-Oax bonding and will be shielded from
an incoming substrate molecule by the equatorial ligands.
On the other hand, the β-dxz,yz, which are also low-lying
empty orbitals, are expected to be less important as
acceptor orbitals than the R-dz2 (σ*), since they are more
than 0.7 eV higher in energy. Nevertheless, involvement
of these π orbitals, particularly the lowest one, the β-dxz,
is a distinct possibility.
and Fe-O4 bonds, to the result that in 1 all four the
equatorial Fe-O bonds have a comparable length. As can
be inferred from the diagrams of Figure 2 the quintet and
triplet electronic configurations differ by a spin-forbid-
den single ligand field transition from the R-dx2-y2 to the
β-dxy. Since the former is strongly σ-antibonding with the
equatorial ligands, but both are (nearly) perpendicular to
the Fe-Oax bond, it becomes evident why the Fe-Oax
bond does not change in going from the quintet to the
triplet spin state, but the equatorial ligands bind much
more strongly to the central iron in the triplet than in the
quintet. In the isomer 2 two of the three equatorial Fe-O
bonds involving the oxygen atoms bonded to the Al
atoms, Fe-O2 and Fe-O3, shorten significantly, whereas
the one involving the oxygen atom bonded only to Si
atoms elongates. The lengthening of the Fe-OSi bond
and the shortening of the trans Fe-O2 bond are parti-
cularly pronounced at the OPBE level. This is consis-
tent with the well-known tendency of this functional to
exaggerate the length of weak bonds trans to strong
bonds.66,67,79
5
5
Triplet Species 31 and 32. The bond lengths gathered in
Table 3 for the S=1 isomers show that, while the Fe-Oax
distance changes only marginally going from the quintet
to the triplet state, the equatorial Fe-O bonds undergo
significant changes. In 1 the equatorial Fe-O bonds ex-
perience a sensible contraction, specially the trans Fe-O2

3874 Inorganic Chemistry, Vol. 49, No. 8, 2010
Rosa et al.
Figure 7. DFT/ZORA/BP86/TZ2P potential energy profiles for methane hydroxylation reaction by 3,51. Energies are in kcal/mol and include ZPE
correction.
Scheme 1. Schematic Representation of the Rebound Mechanism
From the diagrams of Figures 5 and 6 it is apparent that
the most striking difference between the quintet and
triplet spin states is the reduced splitting of the R and
β levels in the latter. This is because the lowering of the
spin multiplicity from S=2 to S=1 makes the exchange
field of the unpaired R-spin electrons less stabilizing. With
regard to the reactivity, worth mentioning are the upshift
of the R-dz2 (σ*) in the triplet, which then will less readily
accept electrons from the C-H σ orbital, and the reduced
gap between the R-dz2 (σ*) and the unoccupied β-spin
Fe-Oax π* orbitals. Actually the lowest β-spin Fe-Oax
π* orbital, the dxz, is nearly degenerate with the R-dz2 (σ*)
3
3
in 1 and only 0.15 eV higher in 2, thus opening a π
reaction pathway in the H-abstraction from methane.
Methane Hydroxylation by the Fe-ZSM-5 “r-oxygen”
Model. In this section we present the results for the methane
hydroxylation by the zeolite “R-oxygen” model, ZFeO,
following the rebound mechanism shown in Scheme 1. This
involves initial hydrogen abstraction from methane by
ZFeO to yield ZeFeOH together with a methyl radical
(intermediate I), followed by methyl radical rebound on
the ferric-hydroxo intermediate, I, to generate the ferric-
methanol complex, PC, which then releases the methanol
and restores the unoxidized ZFe cluster.
˚
Figure 8. DFT/ZORA/BP86/TZ2P optimized geometries (A and deg-
rees) of the species involved in the H-abstraction from methane by 3,51.
The reaction path for the formation of methanol from
methane and ZFeO was explored for the quintet and triplet
spin states of isomer 1 and for the quintet spin state of
isomer 2, the triplet state of this isomer being too high in
energy to play a role in the methane hydroxylation reaction.
First, we will focus on the reactivity of isomer 1 in the
quintet ground spin state and in the triplet excited state.
Then, we will compare the reactivity of the isomers 1 and
2 in their quintet ground spin state.
Figure 7 shows the potential energy profiles for the
methane hydroxylation reaction by 3,51 computed at the
DFT/ZORA/BP86/TZ2P level of theory. The relevant
energy data computed for the methane hydroxylation
reaction by 3,51 and 52 are collected in Table 3.
The key geometrical features of the species involved in
the methane hydroxylation reaction by 3,51 are shown in
Figures 8 and 9, and in Figures S1-S4 in the Supporting

Article
Inorganic Chemistry, Vol. 49, No. 8, 2010 3875
geometry it is evident that not only the R-dz2 (σ*) but also
the β-dxz orbital is involved as acceptor orbital: the 145.5ꢀ
and 139.6ꢀ computed for the —FeOH angle at the BP86
and OPBE level, respectively, clearly indicate that the
methane σC-H donor orbital seeks a compromise position
between linear (180ꢀ), to optimize the overlap with the
R-dz2 (σ*), and 90ꢀ, to optimize the overlap with the β-dxz
MO (the 2πx* in Figure 1 is nominally the β-dxz MO; it is
in fact a Fe-dxz-O-2px antibonding orbital, the σC-H
donor orbital can maximize its overlap with the O-2px by
decreasing the angle to 90ꢀ).19,20 This is different from the
TS for the pentaaqua complex (H2O)5 FeO2þ(S = 2),
where the TS is linear (—FeOH angle ca. 180ꢀ), suggest-
ing almost exclusive involvement of the R-dz2 (σ*). This
predominance of the R-dz2 (σ*) and the linear TS cannot
be explained completely by the lower energy of the R-dz2
(σ*): it is lower than 2πx*, but only by 0.28 eV (cf.
Figure 1). Other factors—overlap of σ(C-H) with the
R-dz2 (σ*) and 2πx* respectively, and steric repulsion
effects—are also important in determining the relative
importance of the σ* and π* channels.19,20
5
Relaxation of TSH toward products leads to inter-
mediate I, which has a Fe-bound hydroxo group (dFe-OH=
˚
1.767 A) weakly interacting with the methyl carbon
˚
(dOH C = 1.767 A). The high spin density found on the
3 3 3
methyl carbon (-0.87) indicates substantial radical char-
acter in the CH3 fragment. The spin densities on the iron and
axial oxygen atoms are also high, higher than in the transi-
tion state. At the DFT/ZORA/BP86/TZ2P level of theory
5
5
5
the H-abstraction reaction (CH4 þ 1 f TSH f I) has
quite a low energy barrier of 6.6 kcal/mol and is endother-
mic by 6.5 kcal/mol. Comparable results are obtained at
different levels of theory, as inferred from the ΔETqS and
ΔE1 values gathered in Table 3. It is gratifying that thHe two
different codes, ADF (STO basis functions) and Turbomole
(Gaussian basis functions), give quite close results, and the
same holds for the functionals BP86 and OPBE. Only in the
final step toward the products the OPBE functional clearly
gives a significantly larger lowering of the total energy.
Intermediate I undergoes OH-rebound generating the pro-
duct complex, 5PC. During this step the hydroxyl group of
the intermediate complex is transferred to the methyl car-
bon, yielding methanol. In the product complex the methyl
alcohol remains coordinatively bound to the metal with the
˚
Fe O distance ranging from 2.059 to 2.118 A, depending
3 3 3
on the functional and basis set used (see Figure 9 and
Figures S3 and S4 in the Supporting Information). In 5PC
the hydroxyl hydrogen is engaged in a rather strong hydro-
gen bond with the closest oxygen atom of the zeolite
framework, which induces significant changes in the iron
coordination sphere. There is no evidence of such an
intramolecular hydrogen bond interaction in the 5PC struc-
ture obtained at the OPBE level (see Figure S4 in the
Supporting Information), which is consistent with this
functional failing to describe hydrogen bonds.72 The energy
data in Table 3 show that the rebound step is highly exo-
thermic, ΔE2 ranges from -38.7 to -44.5 kcal/mol, and
has a very low energy barrier of ∼ 2 kcal/mol, indicating
that OH-rebound is, as always, much faster than H-abstrac-
tion. We estimated the KIE for 5TSH using the approximate
Eyring expression based on ZPEs only. Substitution of all
hydrogen atoms of methane by deuterium leads to KIE
values ranging from 5.4 to 5.7, depending on the level of
˚
Figure 9. DFT/ZORA/BP86/TZ2P optimized geometries (A and deg-
rees) of the species involved in the rebound phase of methane hydroxyla-
tion by 3,51.
Information. The spin densities and charges computed
for these species are reported in Table 4.
We consider first the methane hydroxylation path on
the quintet spin surface, starting with H-abstraction. This
step involves attack of methane on the catalyst leading to
5
the transition state, TSH, without formation of a reac-
tant complex, that is, a complex in which the oxo of ferryl
is weakly bound to methane. The geometrical parameters
in Figure 8 show that in 5TSH the transferring hydrogen is
closer to the oxygen atom than to the carbon (dO 3 3 3
=
H
˚
˚
1.15 A and dC 3 3 3 H=1.36 A), which means we are dealing
with a rather late transition state. The spin density
distribution in the TS shows already a quite pronounced
radical character of the methyl. From the transition state

3876 Inorganic Chemistry, Vol. 49, No. 8, 2010
Rosa et al.
Table 4. Atomic Spin Densities (ζ) and Charges (q) Computed at Different Levels of Theory for the Reaction Intermediates in the Rebound Mechanism of Methane-to-
Methanol Oxidation by 3,51
TSH
BP86a/BP86b
I
TSreb
PC
BP86a/BP86b
P
OPBE
BP86a/BP86b
OPBE
BP86a/BP86b
OPBE
BP86a/BP86b
OPBE
51
ζFe
ζO
ζC
ζH
qFe
qO
qC
qH
3.88/3.86
0.28/0.34
-0.51/-0.54
-0.02/-0.01
1.09/0.80
-0.50/-0.55
0.59/-0.44
0.15/0.31
3.97
0.26
-0.51
-0.03
1.19
-0.55
0.60
0.20
4.07/4.03
0.38/0.44
-0.87/-0.89
-0.01/-0.02
1.11/0.77
-0.49/-0.57
0.47/-0.44
0.18/0.33
4.09
0.28
-0.69
-0.03
1.20
-0.55
0.57
0.21
4.08/4.05
0.40/0.45
-0.92/-0.93
0.01/0.00
1.12/0.75
-0.49/-0.56
0.46/-0.40
0.15/0.32
3.77/3.76
0.04/0.04
0.00/0.00
0.00/0.00
0.93/0.65
-0.53/-0.43
0.90/-0.23
0.28/0.37
3.82
0.02
0.00/0.00
0.00/0.00
1.04
-0.56
0.89
0.25
3.74/3.77
0.00/0.00
0.00/0.00
0.00/0.00
0.88/0.60
-0.51/-0.44
0.84/-0.18
0.21/0.30
3.78
0.00
0.00
0.00
0.91
-0.51
0.80
0.23
31
ζFe
ζO
ζC
ζH
qFe
qO
qC
qH
2.51/2.52
0.04/0.06
-0.65/-0.69
-0.02/-0.01
0.96/0.62
-0.46/-0.47
0.55/-0.44
0.18/0.33
2.62
2.58/2.59
0.11/0.13
-0.82/-0.86
-0.02/-0.02
0.97/0.62
-0.45/-0.47
0.48/-0.43
0.19/0.33
2.73
0.11
-1.00
-0.03
0.98
-0.46
0.44
0.24
2.59/2.58
0.20/0.22
-0.94/-0.95
0.00/0.00
1.00/0.60
-0.45/-0.48
0.45/-0.42
0.15/0.33
2.01/2.01
0.04/0.04
0.00/0.00
0.00/0.00
0.84/0.51
-0.51/-0.41
0.90/-0.23
0.26/0.34
2.07
0.02
0.00/0.00
0.00/0.00
0.88
-0.54
0.91
0.23
2.04/2.09
0.00/0.00
0.00/0.00
0.00/0.00
0.77/0.44
-0.51/-0.44
0.84/-0.18
0.21/0.30
2.08
0.00
0.00
0.00
0.78
-0.51
0.80
0.23
-0.03
-0.67
-0.03
0.97
-0.47
0.56
0.26
a ADF (ZORA/TZ2P). b TURBOMOLE (def2-TZVP).
theory, at a temperature of 298.15 K. The calculated KIE
values compare well with the ones obtained experimentally
for methane versus methane-d2 oxidation by “R-oxygen”
for which, depending on the temperature, KIE values
ranging from 1.9 to 5.5 were measured.34
To obtain the transition state geometry of the rebound
step, TSreb, we have first performed a potential energy
scan along the intrinsic reaction coordinate, that is, the
—HOC angle, the one describing the rotation of the
methyl radical around the hydroxyl hydrogen to bind
the oxygen.12 The maximum energy structure was then
taken as a guess for the transition state search. These
calculations were performed using the BP86 functional.
We were not able to perform a potential energy scan at the
OPBE level because no structure could be optimized.
5TSreb has a reactant-like structure with a short Fe-OH
provided at least in part, by the methane hydroxylation
reaction itself which is highly exothermic on the quintet sur-
face, ΔE ranges from -10.9 to -20.1 kcal/mol (see Table 3).
Desorption of the methanol affords the ZFe cluster in
its quintet ground spin state and the metal in the 2þ
oxidation state. The molecular and electronic structure
data computed for this complex agree well with those
previously reported by Kachurovskaya et al.44
Coming now to the reactivity on the low-spin surface,
we note that the triplet surface lies above the quintet
surface in the entrance channel of the oxidative process,
especially at the OPBE level (see Table 2), and is highly
destabilized during the complete reaction process thus
making spin crossover largely unfavorable energetically.
This is immediately apparent in Figure 7. The energy data
in Table 3 show that the H-abstraction process on the
triplet surface is much more endothermic, ΔE1 ranges
from 17.3 to 17.8 kcal/mol, and has a much higher energy
barrier (more than 18 kcal/mol) than on the quintet
˚
distance of 1.796 A, long C O and C H distances of
3 3 3
3 3 3
˚
2.760 and 2.169 A, respectively, and a —HOC angle of
44.4ꢀ. Notably, the carbon atom does not lie on the FeOH
plane, the —FeOHC dihedral angle being 134ꢀ. The
displacement of the carbon atom out of the FeOH plane
serves to maximize the overlap between the carbon 2pz
and the pertinent oxygen lone pair.
3
surface. On account of the higher energy barrier, TSH
occurs later than TSH. According to the geometrical
5
parameters in Figure 8 and Figures S1 and S2 in the
Supporting Information, in TSH the O H distance is
3
3 3 3
5
shorter and the C H distance is longer than in TSH
The reaction product, methanol, is released in a sub-
sequent dissociative step, 5PC f 5ZFe þ CH3OH, with an
energy cost of 23.8 kcal/mol (24.5 kcal/mol at the BP86/
def2-TZVP level) relative to PC. At the OPBE level this
step is less endothermic and costs 12.0 kcal/mol (this is the
only instance of a significant difference between the BP86
and OPBE functionals in Table 3). Anyhow, the energy
required for the desorption of methanol is quite high, which
is in agreement with the statement of Panov et al.34,87 that
methanol is strongly bound to the zeolite surface and can
be extracted only by using a mixture of water and aceto-
nitrile. The energy required for methanol desorption can be
3 3 3
˚
˚
(1.10 vs 1.15 A and 1.48 vs 1.36 A, at the DFT/ZORA/
BP86/TZ2P level). This is also reflected in the spin density
distribution showing a more pronounced radical charac-
ter of the methyl in 3TSH. Comparison of the spin density
3
3
distribution in TSH and 1 reveals that the transition
state species involves a significant charge transfer from
the methane toward the ferryl moiety. The charge transfer
involves both the Fe-O σ antibonding MO, the R-dz2
(σ*), and the Fe-O π antibonding MO, the β-dxz. This is
not surprising as these acceptor orbitals are nearly degen-
erate in 31 (see diagram of Figure 5). In 3TSH the —FeOH
angle is 121ꢀ, much smaller than in 5TSH, indicating that
the transition state geometry is largely controlled by the
interaction of the C-H σ bond of the substrate with the
(87) Sobolev, V. I.; Dubkov, K. A.; Panna, O. V.; Panov, G. I. Catal.
Today 1995, 24, 251.

Article
Inorganic Chemistry, Vol. 49, No. 8, 2010 3877
Figure 10. DFT/ZORA/BP86/TZ2P potential energy profiles for methane hydroxylation reaction by 52. Energies are in kcal/mol and include ZPE
correction.
β-dxz (2πx*) although there is also a significant R-σC-H
donation into the R-dz2 (σ*). As discussed in detail by
Michel and Baerends20 the donation from the σC-H into
the lowest vacant acceptor orbitals of the catalyst also
controls the H-abstraction energy barrier: the lower the
acceptor orbitals lie in energy the lower ΔETqS is.16,18
Thus, the huge increase of the H-abstractionH energy
barrier observed on going from the quintet to the triplet
surface can be mainly traced to the difference in the
energy of the acceptor orbitals of the electrophile. Be-
cause of the reduced R-exchange field, the R-dz2 (σ*)
acceptor MO lies in 31 at much higher energy than in 51.
Also the 2πx* (β-dxz) and 2πy* (β-dyz) are higher in 31, but
the difference is much less extreme for these dπ* orbitals.
In this case the difference is not due to a difference in
exchange field (the number of β-spin electrons actually
increases), but the effect is indirect: because of the deple-
tion of the antibonding R-dx2-y2, the FeOeq bond dis-
tances contract (the effect is particularly large for Fe-O2
and Fe-O4, see Tables 1 and 2), and this drives the empty
dπ* orbitals up.
ranging from 5.5 kcal/mol (DFT/ZORA/OPBE/TZ2P)
to 15.3 kcal/mol (DFT/BP86/def2-TZVP). The more
facile desorption of methanol from 3PC is consistent with
the methanol being in PC less bound than in PC, as
pointed out above.
According to the values in Table 3 the overall methane
hydroxylation reaction is exothermic also on the triplet
surface, although much less than on the quintet surface.
Our results are at variance with previous theoretical
results by Yoshizawa et al.43 according to which the
methane hydroxylation reaction by “R-oxygen” of Fe-
ZSM-5 zeolite is highly endothermic. However, the re-
sults of these authors are for [FeO]þ instead of [FeO]2þ as
the active catalytic center, which, as pointed out before, is
expected to give much higher barriers.
3
5
Finally, we examine the reactivity of isomer 2 on the
quintet ground state surface. Figure 10 shows the poten-
tial energy profile for the methane hydroxylation reaction
5
by 2 computed at the DFT/ZORA/BP86/TZ2P level.
The key geometrical features of the species involved in the
reaction path are displayed in Figures 11 and 12 and in
Figures S5 and S6 in the Supporting Information. The
spin densities and charges computed for these species are
reported in Table 5.
3
Relaxation of TSH toward products leads to inter-
mediate I. The intermediate complex has a Fe-bound
3
˚
hydroxyl group (dFe-OH = 1.715 A) weakly interacting
˚
with the methyl carbon (dOH C=1.683 A), which shows
The energy profiles for the methane hydroxylation
almost a fully developed spi3n3 3(ζC = -0.82). Subsequent
OH-rebound through TSreb leads to the product com-
reaction by 2 and 1 are very similar. The reaction
starts with the attack of methane on the catalyst leading
to the transition state, 5TSH (2), without formation of a
5
5
3
3
plex, PC. The rebound step is highly exothermic (ΔE2
5
ranges from -34.4 to -39.3 kcal/mol) although less than
in the case of the quintet spin state. This is because 3PC is
less stable than 5PC because of a less effective interaction
reactant complex. Relaxation of TSH (2) toward pro-
5
ducts leads to intermediate I (2). From the ΔE1 and
ΔETS q values gathered in Table 3 it is apparent that the
H
H-abstraction step is more facile for 52 than for 51, both
at the thermodynamic and kinetic levels. On account of
between methanol and iron in the former. The Fe
distance is indeed in PC longer than in PC (2.136 vs
O
3 3 3
3
5
5
the lower energy barrier, TSH (2) occurs earlier than
˚
2.059 A, at the DFT/ZORA/BP86/TZ2P level). The re-
bound step is also on the low spin surface very fast
5TSH (1).
(ΔETqS 3.5-3.37 kcal/mol), although the barrier is a
The geometrical parameters in Figure 11 and Figures
S5 and S6 in the Supporting Information show, indeed,
reb
little higher than on the quintet surface. The important
point here is that regardless of the spin state OH-rebound
is much more facile than the H-abstraction, both at the
thermodynamic and kinetic levels. As can be inferred
from Figure 9 and Figure S3 in the Supporting Informa-
tion, the salient structural features of 3TSreb do not differ
significantly from those of 5TSreb save for a larger —HOC
angle (50.0-53.2ꢀ) and a smaller —FeOHC dihedral
angle (111.8-114.4ꢀ).
that in 5TSH (2) the transferring hydrogen is almost
˚
midway between C and O (dC 3 3 3 H = 1.292 A, dO 3 3 3
=
H
˚
1.210 A, at the DFT/ZORA/BP86/TZ2P level). That
5TSH (2) occurs earlier than 5TSH (1) is also evident from
the spin density distribution in Table 5 showing a less
pronounced radical character of the carbon atom of the
methyl group in the former. Similar to 5TSH (1), 5TSH (2)
settles for a compromise geometry that allows interaction
of the methane σC-H donor orbital with both the acceptor
MOs of the catalyst by assuming an —FeOH angle of
The release of methanol from 3PC is energetically more
favorable than from 5PC as it requires an energy amount

3878 Inorganic Chemistry, Vol. 49, No. 8, 2010
Rosa et al.
˚
Figure 11. DFT/ZORA/BP86/TZ2P optimized geometries (A and
degrees) of the species involved in the H-abstraction from methane by 52.
153.7ꢀ (141.8ꢀ at the OPBE level), very similar to the
145.5ꢀ for 5TSH (1).
The small lowering of the H-abstraction barrier going
5
5
from 1 to 2, from 6.6 to 5.2 kcal/mol, and the subtle
molecular and electronic structure differences between
5TSH (1) and 5TSH (2) as well, can be traced primarily to
small differences in energy and composition of the lowest
acceptor MOs in the two isomers. As a matter of fact, the
main acceptor MO, the R-dz2 (σ*), is in 52 ∼ 0.2 eV lower
than in 51. In turn, the β-dxz lies at about the same energy
5
in the two isomers, but has in 2 a higher oxo character
than in 51 (39% vs 36%).
The change of the metal coordination mode has a
very minor influence on the rebound step, as can be
inferred from the ΔETqS and ΔE2 values reported in
reb
Table 3 for 51 and 52. Thus, H-abstraction is also for 52
the rate-determining step. Also the KIE value esti-
˚
Figure 12. DFT/ZORA/BP86/TZ3P optimized geometries (A and
degrees) of the species involved in the rebound phase of methane hydro-
xylation by 52.
5
mated for TSH (2) is not so different from the one
5
computed for TSH (1). Substitution of all hydrogen
atoms of methane by deuterium leads, indeed, to a KIE
value ranging from 4.9 to 5.1, depending on the level of
theory. Notably, the KIE value computed for methane
versus methane-d2 oxidation, which is more compar-
able to the experiment,34 is 4.7 at the DFT/ZORA/
BP86/TZ2P level.
case of 51 (-7.8 vs -10.9 kcal/mol, at the DFT/ZORA/
BP86/TZ2P level).
In summary, our theoretical results suggest that the
rebound mechanism is a feasible pathway for methane
oxidation by ZFeO. The calculations show that the
triplet state of the oxo species, 31, can hardly be involved
in C-H activation. By contrast, the reaction proceeds
with very low barriers on the quintet surface of either 1
or 2 isomers, although the barrier for H-abstraction,
which is the rate-determining step, is slightly lower for
the isomer 2.
5
Release of methanol from PC (2) is energetically
more demanding than from 5PC (1) (27.1 vs 23.8 kcal/
mol, at the DFT/ZORA/BP86/TZ2P level). This is
5
5
because ZFe (2) is slightly less stable than ZFe (1).
For the same reason the overall methane hydroxylation
reaction is less exothermic in the case of 52 than in the

Article
Inorganic Chemistry, Vol. 49, No. 8, 2010 3879
Table 5. Atomic Spin Densities (ζ) and Charges (q) Computed at Different Levels of Theory for the Reaction Intermediates in the Rebound Mechanism of Methane-to-
Methanol Oxidation by 52
TSH
BP86a/BP86b
I
TSreb
PC
BP86a/BP86b
P
OPBE
BP86a/BP86b
OPBE
BP86a/BP86b
OPBE
BP86a/BP86b
OPBE
52
ζFe
ζO
ζC
ζH
qFe
qO
qC
qH
3.77/3.77
0.30/0.33
-0.39/-0.44
-0.03/-0.02
1.07/0.84
-0.50/-0.52
0.61/-0.45
0.12/0.30
3.90
0.25
-0.43
-0.04
1.17
-0.55
0.61
0.17
4.05/4.00
0.37/0.43
-0.83/-0.86
-0.02/-0.02
1.09/0.81
-0.50/-0.56
0.46/-0.43
0.17/0.33
4.15
0.41
-1.00
-0.02
1.21
-0.54
0.43
0.20
4.05/4.00
3.78/3.77
0.03/0.03
0.00/0.00
0.00/0.00
0.93/0.73
-0.54/-0.45
0.90/-0.21
0.28/0.37
3.83
0.01
0.00/0.00
0.00/0.00
1.01
-0.56
0.88
0.28
3.72/3.74
0.00/0.00
0.00/0.00
0.00/0.00
0.91/0.71
-0.51/-0.44
0.84/-0.18
0.21/0.30
3.76
0.00
0.00
0.00
0.92
-0.51
0.80
0.23
0.39/0.42/
-0.85/-0.86
0.01/0.00
1.09/0.78
-0.51/-0.54
0.49/-0.39
0.17/0.33
a ADF (ZORA/TZ2P). b TURBOMOLE (def2-TZVP).
IR Spectra of Intermediate Species. Panov et al.88
postulated the formation of FeOCH3 species on the basis
of the IR spectrum obtained by treatment of (activated)
zeolite with methane. The IR spectrum shows in the
C-H stretching region four bands at 2823, 2919, 2945,
and 2964 cm-1, the last two very broad and weak.
(isomer 2) complex (see Figure 12) shows instead four
bands in the C-H stretching region: 2947, 3027, 3050,
and 3064 cm-1 with the last two remarkably more intense
than the first two. An analysis of these normal modes
reveals that the first three vibrations are associated to the
C-H stretching modes of the methanol in PC, the fourth
is associated to the stretching of the O-H group; this
stretching is much red-shifted relative to the bare metha-
nol (3681 cm-1) because of the weakening of the O-H
bond caused by the hydrogen bonding with the oxygen
atom of the zeolite, see Figure 12.
In conclusion, the C-H stretching region of the experi-
mental IR spectrum does not unequivocally prove the
existence of an FeOCH3 moiety, but it may arise from the
PC of the rebound mechanism, or from contributions of
both PC and ZFeOCH3.
In the same spectrum in the O-H stretching region an
intense band at 3674 cm-1 was observed, which led these
authors to postulate the formation of a FeOH species.
It is important to realize that formation of these two
species can be easily rationalized if the intermediate
complex formed after H-abstraction (5I) first breaks
•
down into ZFeOH and free CH3 . The methyl radical
then may attack another R-site collapsing into the meth-
oxy species. Dissociation of the intermediate is endother-
mic, the amount of energy required depending on the
system at hand. In our case it costs, at the BP86/TZ2P
level, upon ZPE inclusion, 4.8 kcal/mol for the isomer 2.
This value is obtained considering ZFeOH in the sextet
(S=5/2) ground spin state antiferromagnetically coupled
Conclusions
We conclude that the reactivity of the R-oxygen site of Fe-
ZSM-5 can be fully explained as a consequence of an optimal
ligand environment for the [FeO]2þ species offered by the
cation exchange site of the zeolite. Notably, it offers an
equatorial set of oxygen donor ligands, which is weak enough
to keep the Fe in the high-spin configuration. This is
advantageous for a low energy of the R spin LUMO, the
R-dz2 (σ*), which is, with oxygen donor ligands, the most
important acceptor orbital. In the second place, there is the
rather unique circumstance in this particular site for FeO2þ
that it does not have a trans axial donor ligand. This is in
conformance with the ideal situation as proposed in ref 16
which is approximately realized with the coordination of
•
to the S=-1/2 CH3 radical (if one considers ZFeOH in
the energetically higher S = 3/2 spin state coupled to a
•
parallel S = 1/2 CH3 radical the dissociation reaction
costs 20.0 kcal/mol). In turn, the reaction ZFeO (52) þ
•
CH3 f ZFeOCH3 (S = 5/2) is exothermic by 73.3 kcal/
mol (ZPE included). (N. B.: in ZFeOCH3 the Fe-O-C
angle is 148ꢀ). Thus, the intermediate complex may evolve
toward dissociation (4.8 kcal/mol) or undergo oxygen
rebound (TS barrier = 2.0 kcal/mol), the latter process
being slightly favored energetically.
We have also computed the IR spectra of ZFeOH (S=
5/2) and ZFeOCH3 (S = 5/2). For ZFeOH we find an
O-H stretching vibration at 3603 cm-1, that is, suffi-
ciently close to the 3674 cm-1 value experimentally found
to support the experimental assignment. For ZFeOCH3,
we find only 3 IR active C-H vibration modes around
2900 cm-1, at 2888, 2967, and 2971 cm-1, of very weak
and comparable intensity. Although the fourth experi-
mental band at 2945 cm-1 could well be an overtone
band, as suggested by Panov et al.,88 the correspondence
with the experiment is not so good in the case of ZFeO-
CH3. Interestingly, the IR spectrum computed for the PC
EDTA to FeO2þ 16
.
We have given arguments to favor the d4 Fe(IV) metal
state, that is, a FeIVO2þ moiety, as active center. This is at
variance with the interpretation of experimental data from
23
€
Mossbauer spectra and from resonant inelastic X-ray
scattering (RIXS),49,50 which favor a Fe3þ metal, that is, a
quasi-radical [FeIIIO•-]2þ electronic structure. We wish to
caution, however, that the iron charge and the oxygen radical
character of the FeO2þ moiety are very difficult to establish
unequivocally. In a sense we always have radical character,
since the systems are open shell anyway. It is from the
calculations impossible to say whether the electronic struc-
ture is better described as [Fe(III)O•-]2þ than as [FeIV-
(88) Panov, G. I.; Dubkov, A.; Paukshtis, Y. A. In Catalysis by Unique
Metal Ion Structures in Solid Matrices - NATO Science Series, II. Mathe-
matics, Physics and Chemistry; Centi, G., Wichterlova, B., Bell, A. T., Eds.;
Kluwer Academic Publishers: Dordrecht, 2001; Vol. 13; pp 149.
O2- 2þ
are strong mixtures.10 The calculations do show considerable
] , since the orbitals are not purely Fe-3d or O-2p, but

3880 Inorganic Chemistry, Vol. 49, No. 8, 2010
Rosa et al.
delocalization of the spin onto the oxo oxygen, which is of
course far from the O2- electronic configuration of the
formal FeIVO2- picture. The Fe in the orbital scheme of
Figure 1 may be formally FeIV, but of course the charge on Fe
in the complexes is far from þ4 (more like þ1, see Tables). It
is possible that the experimental results can be explained just
as well on the basis of our proposed electronic configuration
as displayed in Figure 1. The crucial point is that a
[Fe(III)O•-]2þ picture for the electronic configuration has
led to the assumption of occupation of the 3σ*, compare
Figure 2 of ref 48 which is definitely contrary to our under-
standing of how FeO2þ acts in oxidation catalysis. Notably, a
d5 Fe(III) high spin electronic configuration, as is implied in
the [Fe(III)O•-]2þ description of the electronic structure, and
which would be present in the FeOþ moiety, would necessa-
rily have the R-dz2 (σ*) orbital occupied, inhibiting its action
as acceptor orbital. It therefore should exhibit much higher
barriers, that is, lower reactivity. This is borne out by
calculations.
hydroxylation reactions. In all cases the rebound mecha-
nism52,53,89 appears to be operative. It is interesting to note
that this mechanism is in agreement with the experimental
results on the simultaneous production of FeOH and
FeOCH3 upon reaction of methane with the R-oxygen sites
by Panov et al.,88 since there is evidence that these are often
produced from binuclear Fe oxygen configurations. The
•
CH3 radical formed in the H-abstraction step may react in
the rebound step with the formed FeOH, but also with the
nearby FeO2þ to form FeOCH3.
Acknowledgment. A.R. and G.R. acknowledge finan-
ꢀ
cial support by the Universita della Basilicata, Italy (RIL
Funds 2006). E.J.B. was supported by WCU (World
Class University) program through the Korea Science
and Engineering Foundation funded by the Ministry of
Education, Science and Technology (Project No. R32-
2008-000-10180-0) and by the Dutch National Research
School Combination “Catalysis by Design”.
The identification of the high-valent [FeIVO2- 2þ
moiety as
]
Supporting Information Available: Additional details are
given in Figures S1-S8. This material is available free of charge
via the Internet at http://pubs.acs.org.
the active species also in the R-oxygen site of Fe-ZSM-5
brings this system in line with the well-known and much
studied biological systems where the ubiquitous [FeIVO2- 2þ
]
moiety appears to be the key active species in oxidation and
(89) Groves, J. T. J. Inorg. Biochem. 2006, 100, 434.