Oxidation of ethane to ethanol by N2 O in a metal-organic framework with coordinatively unsaturated iron(II) sites

Article abstract of DOI:10.1038/nchem.1956

Enzymatic haem and non-haem high-valent iron-oxo species are known to activate strong C-H bonds, yet duplicating this reactivity in a synthetic system remains a formidable challenge. Although instability of the terminal iron-oxo moiety is perhaps the foremost obstacle, steric and electronic factors also limit the activity of previously reported mononuclear iron(IV)-oxo compounds. In particular, although nature's non-haem iron(IV)-oxo compounds possess high-spin S = 2 ground states, this electronic configuration has proved difficult to achieve in a molecular species. These challenges may be mitigated within metal-organic frameworks that feature site-isolated iron centres in a constrained, weak-field ligand environment. Here, we show that the metal-organic framework Fe₂ (dobdc) (dobdc^4- = 2,5-dioxido-1,4- benzenedicarboxylate) and its magnesium-diluted analogue, Fe_0.1 Mg_1.9 (dobdc), are able to activate the C-H bonds of ethane and convert it into ethanol and acetaldehyde using nitrous oxide as the terminal oxidant. Electronic structure calculations indicate that the active oxidant is likely to be a high-spin S = 2 iron(IV)-oxo species.

Full text of DOI:10.1038/nchem.1956

ARTICLES
PUBLISHED ONLINE: 18 MAY 2014 | DOI: 10.1038/NCHEM.1956
Oxidation of ethane to ethanol by N2O in a
metal–organic framework with coordinatively
unsaturated iron(II) sites
Dianne J. Xiao1, Eric D. Bloch1, Jarad A. Mason1, Wendy L. Queen2, Matthew R. Hudson3,
Nora Planas4, Joshua Borycz4, Allison L. Dzubak4, Pragya Verma4, Kyuho Lee2, Francesca Bonino5,
5
6
5
4
4
`
Valentina Crocella , Junko Yano , Silvia Bordiga , Donald G. Truhlar , Laura Gagliardi ,
Craig M. Brown3,7 and Jeffrey R. Long1,8
*
Enzymatic haem and non-haem high-valent iron–oxo species are known to activate strong C–H bonds, yet duplicating this
reactivity in a synthetic system remains a formidable challenge. Although instability of the terminal iron–oxo moiety is
perhaps the foremost obstacle, steric and electronic factors also limit the activity of previously reported mononuclear
iron(IV)–oxo compounds. In particular, although nature’s non-haem iron(IV)–oxo compounds possess high-spin S 5 2
ground states, this electronic configuration has proved difficult to achieve in a molecular species. These challenges may be
mitigated within metal–organic frameworks that feature site-isolated iron centres in a constrained, weak-field ligand
environment. Here, we show that the metal–organic framework Fe2(dobdc) (dobdc42 5 2,5-dioxido-1,4-benzenedicarboxylate)
and its magnesium-diluted analogue, Fe0.1Mg1.9(dobdc), are able to activate the C–H bonds of ethane and convert it into
ethanol and acetaldehyde using nitrous oxide as the terminal oxidant. Electronic structure calculations indicate that the
active oxidant is likely to be a high-spin S 5 2 iron(IV)–oxo species.
he selective and efficient conversion of light alkanes into analogue. Fe-ZSM-5 (ZSM ¼ Zeolite Socony Mobil), for example,
value-added chemicals remains an outstanding challenge has been shown to oxidize methane to methanol stoichiometrically
T
with tremendous economic and environmental impacts1,2, when pretreated with nitrous oxide13. However, characterization of
especially given the recent worldwide increase in natural gas these materials is nontrivial because of the presence of multiple
reserves3. In nature, C–H functionalization is carried out by iron species, and the nature of the active sites in Fe-ZSM-5
copper and iron metalloenzymes, which activate dioxygen and remains largely a matter of speculation14.
facilitate two- or four-electron oxidations of organic substrates4–7
The use of a metal–organic framework (MOF) to support iso-
through metal-oxo intermediates. Duplicating this impressive reac- lated terminal iron–oxo moieties is a currently unexplored yet
tivity in synthetic systems has been the focus of intense research. In highly promising area of research. The high surface area, permanent
particular, iron(IV)–oxo complexes have now been characterized porosity, chemical and thermal stability, and synthetic tunability
structurally in various geometries (octahedral, trigonal bipyramidal) displayed by many of these materials makes them appealing in
and spin states (S ¼ 1, S ¼ 2), and have proved to be competent this regard. Additionally, MOFs are typically highly crystalline
catalysts for a variety of oxygenation reactions8,9.
with well-defined metal centres suited for characterization by
However, in the absence of a protective protein superstructure, single-crystal and/or powder-diffraction techniques. Furthermore,
terminal iron–oxo species are highly susceptible to a variety of although molecular iron(IV)–oxo complexes generally utilize nitro-
decomposition pathways, which include dimerization to form gen-based chelating ligands, the metal cations in MOFs are often
oxo-bridged diiron complexes, intramolecular ligand oxidation ligated by weaker-field ligands, such as carboxylates and aryl
and solvent oxidation10. Tethering a molecular iron species to a oxides, which are constrained in their coordination position by
porous solid support, such as silica or polystyrene, could potentially the extended framework structure. Thus, in addition to increased
prevent many of these side-reactions. In practice, however, com- stability, terminal oxos in these materials might also have novel
plexes heterogenized in this manner are difficult to characterize electronic properties and reactivity imparted by their unique
using available techniques, and additional problems associated coordination environment.
with steric crowding, site inaccessibility and metal leaching inevita-
Herein we show that the high-spin iron(II) centres within
bly arise11,12. Iron cations can also be incorporated into zeolites, Fe2(dobdc) (dobdc42 ¼ 2,5-dioxido-1,4-benzenedicarboxylate) can
either as part of the framework or at extraframework sites, to activate N2O, most likely forming a transient, high-spin iron(IV)–
produce reactive iron centres that have no direct molecular oxo intermediate, which rapidly reacts to afford Fe2(OH)2(dobdc).
2
1Department of Chemistry, University of California, Berkeley, California 94720-1460, USA, The Molecular Foundry, Lawrence Berkeley National Laboratory,
Berkeley, California 94720-1460, USA, 3Center for Neutron Research, National Institute of Standards and Technology, Gaithersburg, Maryland 20899, USA,
4Department of Chemistry, Chemical Theory Center, and Supercomputing Institute, University of Minnesota, Minneapolis, Minnesota, 55455-0431, USA,
5Department of Chemistry, NIS and INSTM Reference Centres, University of Turin, I-10135 Torino, Italy, 6Physical Biosciences Division, Lawrence Berkeley
National Laboratory, Berkeley, California 94720-1460, USA, 7Department of Chemical Engineering, University of Delaware, Newark, Delaware 19716, USA,
8Materials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720, USA. e-mail: jrlong@berkeley.edu
*
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
© 2014 Macmillan Publishers Limited. All rights reserved.
1

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1956
ARTICLES
Significantly, the magnesium-diluted analogue, Fe0.1Mg1.9(dobdc), metal–N2O adducts are challenging to synthesize because of the
is found to oxidize ethane into ethanol in the presence of N2O poor s-donating and p-accepting properties of the molecule22.
under mild conditions.
Indeed, of the several proposed binding modes, only one (end-on,
h1-N) has been structurally characterized in a molecular complex23.
To establish the coordination mode of N2O in 1, powder neutron-
Results and discussion
Nitrous oxide coordination and activation. Of the three- diffraction data, which are very sensitive to the atomic assignment
dimensional iron(II)-containing MOFs shown to be stable to of O and N, were collected on a sample dosed with various loadings
desolvation15–18, few possess coordinatively unsaturated metal of N2O. At low loadings, the best fit was an average of approximately
centres in a single, well-defined environment. The compound 60% h1-O and 40% h1-N coordination, with Fe–N2O distances of
Fe2(dobdc) (1), also known as Fe-MOF-74 or CPO-27-Fe, is rare 2.42(3) and 2.39(3) Å, respectively. In both cases, a bent Fe–N2O
in this regard, as the hexagonal channels of the framework are angle close to 1208 was observed (see Fig. 1b). Density functional
lined with a single type of square pyramidal iron(II) site (see theory (DFT) studies of N2O-bound 1 using the M06 functional24
Fig. 1a). The high density and redox-active nature of these open show excellent agreement with experiment (see Supplementary
metal sites engender excellent O2/N2 and hydrocarbon separation Fig. 20). Furthermore, these calculations predict the h1-O coordination
properties18,19. However, with respect to the reactivity of the mode to be favoured over the h1-N mode by just 1.1 kJ mol21
framework, only the hydroxylation of benzene into phenol and (see Supplementary Tables 17 and 22). This is consistent with the
the oxidation of methanol into formaldehyde have been nearly equal population split observed, although the magnitude of
reported20,21. Thus, we embarked on a study of its reactivity the difference in energy is smaller than the potential error associated
towards nitrous oxide, a gaseous two-electron oxidant and O- with the calculations.
atom transfer agent that is widely employed in industry,
anticipating the generation of a highly reactive iron(IV)–oxo proposed in a variety of systems ranging from isolated metal atoms
species capable of oxidizing strong C–H bonds.
to iron zeolites22,25,26, h1-N coordination with a bent Fe–N–N angle
Although h1-O coordination with a bent Fe–O–N angle has been
We first investigated the binding of nitrous oxide to 1 under con- is much more unusual. It suggests little p-back-bonding from the
ditions in which the Fe–N2O interaction is reversible. Experimental metal d orbitals into the p* of N2O. This is in contrast to previously
studies on the coordination chemistry of N2O are scarce, as reported vanadium– and ruthenium–N2O adducts, which have
linear metal–N–N–O geometries and for which p-interactions
have been invoked as significant contributors to the stability of
the complexes23,27–29. The bent geometry, long Fe–N2O bond
length and mixed N and O coordination indicate that N2O is
bound only weakly to the iron(II) centres in the framework, a
hypothesis corroborated by in situ transmission-mode infrared
spectroscopy. Spectra collected on a thin film of 1 dosed at
a
room temperature with N2O display a maximum at 2,226 cm21
,
which is very close to the fundamental n(N–N) transition for
unbound N2O (2,224 cm21) and suggests a physically adsorbed
phase with little to no perturbation of the N2O molecule (see
Supplementary Fig. 1). As expected, this interaction is fully
reversible, and the band disappears completely under an applied
vacuum. Consistent with these experimental results, DFT studies
calculate binding energies of 45.6 and 44.5 kJ mol21 for the h1-O
and h1-N modes, respectively, with a natural bond order analysis30
that shows weak back-bonding in both configurations (see
Supplementary Table 23).
On heating the N2O-dosed framework to 60 8C, the material
undergoes a drastic colour change from bright green to dark red–
brown, suggestive of oxidation. In addition, in situ infrared studies
using CO as a probe molecule show that the open metal sites,
which coordinate CO strongly, have been almost entirely consumed
(see Supplementary Fig. 9). Characterization of the resulting
product is consistent with the formulation Fe2(OH)2(dobdc) (2), in
which each iron centre is in the þ3 oxidation state and bound to a
terminal hydroxide anion (see Fig. 2a). Compound 2 is likely to be
formed via a fleeting iron–oxo intermediate, which rapidly undergoes
hydrogen-atom abstraction, although the source of the hydrogen
atom has not yet been determined. Mo¨ssbauer spectroscopy was
used to probe the local environment of the iron centres in the
oxidized material. The 57Fe Mo¨ssbauer spectrum of 2 consists of a
doublet characterized by an isomer shift (d) of 0.40(2) mm s21 and
a quadrupole splitting (|DEQ|) of 0.96(1) mm s21 (see Fig. 2c). The
isomer shift for the iron centres in 2 is similar to the parameters
obtained for the peroxide-coordinated iron(III) centres in
Fe2(O2)(dobdc) (ref. 18), and is consistent with other high-spin
haem and non-haem iron(III) species31–33. In addition, the infrared
spectrum of 2 shows the appearance of two new bands as compared
b
117(2)°
122(2)°
2.42(3) Å
2.39(3) Å
η1-O mode (60%)
η1-N mode (40%)
Figure 1 | Structure of bare and N2O-dosed Fe2(dobdc) (1). a, Structure of
Fe2(dobdc), showing hexagonal channels lined with five-coordinate iron(II)
sites. The view is down the c axis, along the helical chains of iron(II) ions.
b, Experimental structures for N2O binding in Fe2(dobdc), solved from
powder neutron diffraction data collected at 10 K. The molecule binds with a
bent Fe–N2O angle, with a mixture of 60% h1-O coordination and 40% h1-N
coordination (for a comparison of calculated structures with experimental
ones, see Supplementary Fig. 20). Orange, grey, dark blue and red spheres
represent Fe, C, N and O, respectively; H atoms are omitted for clarity.
to the unoxidized framework, which we assign as Fe–OH (667 cm21
)
and O–H (3,678 cm21) vibrations. These bands shift to 639 and
2
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
© 2014 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1956
ARTICLES
a
b
c
N2O
d
II
III
Fe 2(dobdc)
Fe 2(OH)2(dobdc)
O1
100.0
99.5
99.0
98.5
98.0
60 °C
1
2
1.92(1) Å
Fe
O5
O2
O4
O3
3,678
667
3,700 3,650
O6
720
700
680
660
640
620
–4
–2
0
2
4
Wavenumber (cm–1
)
Velocity (mm s–1
)
Figure 2 | Preparation, spectroscopic characterization and structure of Fe2(OH)2(dobdc). a, Reaction scheme for the preparation of Fe2(OH)2(dobdc) (2)
from Fe2(dobdc) (1). b, Infrared spectrum of partially oxidized samples of Fe2(OH)0.6(dobdc) (2′) (black line) and Fe2(18OH)0.6(dobdc) (dotted red line).
The peaks at 667 and 3,678 cm21 shift to 639 and 3,668 cm21, respectively, when N218O is used, which confirms that these are, indeed, new bands derived
from N2O and not simply shifted framework bands. This also strongly supports the assignment of these peaks as Fe–OH and O–H vibrations. c, Mo¨ssbauer
spectrum of 2, with the fit in black. The red component has parameters consistent with high-spin Fe(III) (d ¼ 0.40(2) mm s21, |DEQ| ¼ 0.96(1) mm s21
,
area ¼ 80(2)%). A minor component (green) is assigned as unreacted Fe(II) sites, and another minor component (purple) is assigned as an amorphous
Fe(III) decomposition product. d, The structure of 2 obtained from powder X-ray diffraction data (100 K). The Fe–OH bond distance of 1.92(1) Å is consistent
with that in previously reported Fe(III)–OH compounds. The hydrogen atom on the hydroxide is shown for clarity, but was not found from the diffraction data.
Selected interatomic distances (Å) for 2: Fe–O1 ¼ 1.92(1), Fe–O2 ¼ 2.01(1), Fe–O3 ¼ 2.08(1), Fe–O4 ¼ 2.04(1), Fe–O5 ¼ 2.04(1), Fe–O6 ¼ 2.20(1),
Fe–Fe ¼ 3.16(1). Orange, red and light blue spheres represent Fe, O and H, respectively.
3,668 cm21, respectively, when N218O is employed for the oxidation; Oxidation of ethane to give ethanol. As the isolation of an
the observed differences of 28 and 10 cm21 are very close to the iron(III)–hydroxide product from a reaction that employs a
theoretical isotopic shifts of 27 and 12 cm21 predicted by a two-electron oxidant strongly suggests the intermediacy of an
simple harmonic oscillator model (see Fig. 2b). Partial oxidation of iron(IV)–oxo species, we next carried out the oxidation in the
the framework was achieved by heating at 35 8C for 12 hours, presence of a hydrocarbon substrate that contained stronger C–H
which led to the formation of Fe2(OH)0.6(dobdc) (2′), which bonds, specifically ethane (C–H bond dissociation energy of
has a similar infrared spectrum (although the bands associated 423 kJ mol21), hoping to intercept the oxo species before its
with Fe–OH are less intense) and Mo¨ssbauer parameters (see decay. Indeed, flowing an N2O:ethane:Ar mixture (10:25:65) over
Supplementary Table 9).
the framework at 75 8C led to the formation of various ethane-
The framework maintains both crystallinity and porosity after derived oxygenates, which included ethanol, acetaldehyde, diethyl
oxidation, with a Brunauer–Emmett–Teller (BET) surface area of ether and other ether oligomers, as determined by 1H NMR
1,013 m2 g21 and a Langmuir surface area of 1,171 m2 g21
.
spectroscopy of the extracted products. The formation of ether
Rietveld analysis of powder X-ray diffraction data collected at products is not unprecedented, as N2O-treated Fe-ZSM-5 forms a
100 K on 2 firmly established the presence of a new Fe–O bond, small amount of dimethyl ether in addition to methanol when
but did not reveal whether a hydrogen atom was present. exposed to methane, via a mechanism proposed to involve methyl
However, the Fe–OH bond distance of 1.92(1) Å is consistent radicals as well as multiple iron sites38. We hypothesized that the
with the bond lengths of previously reported octahedral iron(III)– complex mixture of products was related to the close proximity of
hydroxide complexes (1.84–1.93 Å) (see Fig. 2d)34. In addition, reactive iron centres, which are 8.13(2) Å and 6.84(1) Å apart
the trans Fe–Oaxial bond is slightly elongated (Fe–Oaxial ¼ 2.20(1) Å, across and along a channel, respectively, in 1. To avoid
average Fe–Oequatorial ¼ 2.04(1) Å), with the iron centre shifted oligomerization and overoxidation,
a
mixed-metal MOF,
slightly (by 0.23(1) Å) out of the plane of the four equatorial Fe0.1Mg1.9(dobdc) (3), in which the iron(II) sites are diluted with
oxygen atoms. Extended X-ray absorption fine structure (EXAFS) redox-inactive magnesium(II) centres, was synthesized. The BET
analysis of the same sample, as well as periodic DFT calculations, surface area of 1,670 m2 g21 for this material falls between the
provided bond lengths that are consistent with those obtained surface areas of the pure iron and pure magnesium frameworks
from the diffraction data (see Supplementary Table 8).
(1,360 and 1,800 m2 g21, respectively). Although determining the
Surprisingly, the iron(III)–hydroxide species is capable of exact distribution of metal centres in heterometallic MOFs is
activating weak C–H bonds. When the partially oxidized sample challenging, the unit-cell parameters of 3 are between those of
2′ was exposed to 1,4-cyclohexadiene (C–H bond dissociation Fe2(dobdc) and Mg2(dobdc) (see Supplementary Table 10), which
energy of 322 kJ mol21
)
at room temperature, benzene was suggests the formation of a solid solution rather than a mixture of
35
produced as the sole product in quantitative yield. In the process, two separate phases. Additionally, the Mo¨ssbauer spectrum of 3
the framework converted entirely back to iron(II), as determined shows sharp doublets with a significantly different quadrupole
by Mo¨ssbauer spectroscopy. Such reactivity is rare, but not un- splitting to that of the all-iron analogue (2.25(1) mm s21 versus
precedented, for iron(III)–hydroxide compounds. For instance, 2.02(1) mm s21 in Fe2(dobdc); see Supplementary Table 9),
lipoxygenase, an enzyme that converts 1,4-dienes into alkyl which indicates that the iron centres in the magnesium-diluted
hydroperoxides, is believed to proceed through a non-haem ferric framework are in an altered, but uniform, environment. Thus, 3 is
hydroxide intermediate5 and several molecular lipoxygenase possibly best described as containing either isolated iron centres
mimics have also been reported to activate the C–H bond of or short multiiron segments dispersed evenly throughout a
1,4-cyclohexadiene and other 1,4-dienes36,37. However, the oxidizing magnesium-based framework.
power of 2 and 2′ is limited, and no reaction was observed with less-
Exposure of 3 to N2O and ethane under the same flow-through
activated C–H bonds.
conditions yielded the exclusive formation of ethanol and
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
3
© 2014 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1956
ARTICLES
½
½
N
N
O
H
H
H
H
Δ
O
O
O
FeII
1-N2O
FeIV
FeIII
FeII
4
2
1-H2O
H3C CH3
OH
Figure 3 | N2O activation and reactivity of Fe2(dobdc) in the oxidation of ethane and 1,4-cyclohexadiene. Heating of N2O-bound Fe2(dobdc) (1-N2O) to
60 8C results in the formation of a transient high-spin Fe(IV)–oxo species (4), which can react with the strong C–H bonds of ethane. In the absence of a
hydrocarbon substrate, the Fe(IV)–oxo quickly decays via hydrogen-atom abstraction into an Fe(III)–hydroxide (2), which is isolable and well characterized.
This hydroxide species can react with weak C–H bonds, such as those in 1,4-cyclohexadiene, to form benzene and H2O-bound Fe2(dobdc) (1-H2O).
acetaldehyde in a 10:1 ratio, albeit in low yield (60% with respect to quintet spin states. A quintet ground state was predicted, with a
iron). Gas chromatography analysis of the headspace revealed no short Fe–O bond length of 1.64 Å, consistent with that of
ethanol, acetaldehyde or CO, which suggests the products remained previously reported iron(IV)–oxo complexes (see Fig. 4 and
bound to the framework (either at open iron or open magnesium Supplementary Table 11)8. The periodic structure was then
sites), and may explain the high ethanol selectivity. Although the truncated to an 89-atom model cluster41,42 that contained three
framework was still highly crystalline after N2O/ethane treatment, metal centres, six organic linkers and an oxo moiety to facilitate
Mo¨ssbauer spectroscopy revealed that roughly 90% of the iron calculations using more accurate methods. The cluster calculations
centres decayed into a species with similar spectral parameters as were simplified by replacing the two peripheral iron(II) centres
those of 2 (see Supplementary Fig. 18 and Supplementary with closed-shell zinc(II) centres, which have the same charge and
Table 9). We propose that the formation of iron(III)–hydroxide or a similar ionic radius to iron(II) and magnesium(II) cations (see
–alkoxide decay products prematurely halts the catalytic cycle, Supplementary Fig. 19). The geometry of this cluster was then
which leads to substoichiometric yields of hydroxylated product optimized, with all atoms except for the central iron and its first
(see Fig. 3). As glass can be a source of H atoms, the reaction was coordination sphere frozen at the coordinates from the periodic
subsequently repeated in a batch mode, rather than flow-through, PBE þ U optimization. As shown in Table 1, the M06//M06-L24
in a Teflon-lined stainless-steel bomb. This produced both higher calculations also predict
a quintet ground state. Further
yields with respect to iron (turnover number ¼ 1.6) and selectivities calculations were performed with several other exchange-
(25:1 ethanol:acetaldehyde). Though the yield based on ethane correlation functionals, and in each case the ground state was
(roughly 1%) is still too low for practical purposes, this demonstrates found to be a quintet (see Supplementary Tables 11–16). Similar
that the system can, indeed, be modestly catalytic if competing results were obtained when the Zn(II) centres in the 89-atom
substrates are excluded.
cluster were replaced with Mg(II) centres (see Supplementary
Tables 20 and 21).
Electronic-structure calculations. As the high reactivity of the
The electronic structure of the cluster model of 4 was examined
iron–oxo species precluded isolation in both Fe2(dobdc) and its further with single-point multiconfigurational complete active space
magnesium-diluted analogue, electronic structure calculations (CASSCF) calculations followed by second-order perturbation
were performed on Fe2(O)2(dobdc) (4) to gain insight into the theory (CASPT2)43,44. Again, the ground state is predicted to be
geometric and electronic structure of iron–oxo units supported the quintet state (see Table 1 and Supplementary Table 18). Both
within the framework. First, periodic PBE þ U39,40 geometry M06//M06-L and CASPT2//PBE yield a spin density of ꢀ3.7 on
optimizations were performed on 4 for the singlet, triplet and iron, consistent with four unpaired spins localized mainly on the
metal (see Supplementary Tables 13 and 19). Density functional
and CASPT2 calculations were also performed on the cluster
a
O1
b
model of 2; all calculations led to a high-spin sextet ground state
for the iron(III) centre (see Supplementary Tables 11–16 and
Supplementary Table 18).
Although spectroscopic and theoretical studies have long attributed
the reactivity of non-haem enzymatic and synthetic iron(IV)–oxo
z2 (σ*Fe–O
x2 – y2
)
1.64 Å
Fe
O5
O4
E
xz, yz (π*Fe–O
)
O2
O3
Table 1 | Calculated relative energies (kJ mol21) of the
cluster model of 4.
xy
O6
S
M06//M06-L
CASPT2//PBE
0
1
2
210.6
136.4
0.0
249.4
127.6
0.0
Figure 4 | Structure and qualitative MO diagram of Fe2(O)2(dobdc) (4).
a, DFT and CASSCF/PT2 studies predict a short iron–oxo bond (1.64 Å) for
Fe2(O)2(dobdc) (4). Selected interatomic distances (Å) for 4: Fe–O1 ¼ 1.638;
Fe–O2 ¼ 2.004; Fe–O3 ¼ 2.127; Fe–O4 ¼ 2.019; Fe–O5 ¼ 2.054; Fe–O6 ¼
2.140. Orange and red spheres represent Fe and O respectively. b, DFTand
CASSCF/PT2 studies all predict a high-spin S ¼ 2 ground state for
iron(IV)–oxo compounds installed in the Fe2(dobdc) framework.
Density functional and CASPT2 calculations were performed on
a
truncated model of
Fe2(O)2(dobdc) (4) that contained a central iron atom, two peripheral Zn(II) atoms (to replicate
the rigid framework structure) and six organic linkers (see Supplementary Fig. 19). The relative
energies of the cluster models in different spin states are given here. By both methods, the
quintet state (S ¼ 2) is the calculated ground state.
4
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
© 2014 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1956
ARTICLES
are analogous to the 88-atom cluster model of Fe2(dobdc) employed previously41
.
complexes to a quintet spin state45, only a handful of mononuclear
high-spin iron(IV)–oxo species have been characterized46–49, and all
but one exhibit a trigonal bipyramidal coordination geometry50.
In these systems, the oxo moiety is either extremely unstable
([Fe(O)(H2O)5]2þ, for example, has a half-life of roughly ten
seconds) or inaccessible to substrates because of bulky ligand
scaffolds, which lead to sluggish reactivity. However, the
Fe2(dobdc) framework features sterically accessible, site-isolated
metal centres entrenched in a weak-field ligand environment.
Utilizing these two properties, it is possible not only to generate
such a species, albeit fleetingly, but also to direct it towards the
facile activation of one of the strongest C–H bonds known.
The cluster models were simplified further by substituting the two peripheral iron(II)
centres with zinc(II) centres, keeping only the central iron(II) in the cluster.
Constrained geometry optimizations were performed in which only the central iron
and the six oxygen atoms (plus the hydroxide hydrogen in compound 2) of its
first coordination sphere were allowed to relax. Single-point multiconfigurational
complete active space (CASSCF) calculations followed by second-order
perturbation theory (CASPT2) were performed at PBE-optimized geometries of the
cluster models of 2 and 4, and M06 calculations were performed at M06-L
geometries. Full computational details are in the Supplementary Information.
Received 17 December 2013; accepted 14 April 2014;
published online 18 May 2014
References
Concluding remarks
1. Arakawa, H. et al. Catalysis research of relevance to carbon management:
progress, challenges, and opportunities. Chem. Rev. 101, 953–996 (2001).
2. Bergman, R. G. Organometallic chemistry: C–H activation. Nature 446,
391–393 (2007).
The foregoing results demonstrate, through reactivity studies, detailed
characterizations of decay products and theoretical calculations, that
the iron-based MOFs Fe2(dobdc) and Fe0.1Mg1.9(dobdc) are very
likely to be capable of supporting fleeting iron(IV)–oxo species that
possess an unusual S ¼ 2 spin state. With this, Fe2(dobdc) has now
been shown to stabilize iron–superoxo, –peroxo, –hydroxo and
–oxo intermediates, which highlights the promise of MOFs both as
catalysts and as scaffolds for interrogating reactive metal species.
Future work will focus on (1) further exploring the reactivity of
Fe2(dobdc) and its expanded analogues towards ethane and other
hydrocarbon substrates, as well as continued efforts to isolate the
iron-oxo species, (2) the use of dioxygen as the terminal oxidant in
such systems and (3) the design, synthesis and reactivity of other
MOFs with coordinatively unsaturated iron sites.
3. International Energy Agency World Energy Outlook Special Report 2011
http://www.worldenergyoutlook.org/goldenageofgas (2011).
4. Himes, R. A. & Karlin, K. D. Copper–dioxygen complex mediated C–H bond
oxygenation: relevance for particulate methane monooxygenase (pMMO).
Curr. Opin. Chem. Biol. 13, 119–131 (2009).
5. Costas, M., Mehn, M. P., Jensen, M. P. & Que, L. Jr. Dioxygen activation at
mononuclear nonheme iron active sites: enzymes, models, and intermediates.
Chem. Rev. 104, 939–986 (2004).
6. Meunier, B., de Visser, S. P. & Shaik, S. Mechanism of oxidation reactions
catalyzed by cytochrome P450 enzymes. Chem. Rev. 104, 3947–3980 (2004).
7. Wallar, B. J. & Lipscomb, J. D. Dioxygen activation by enzymes containing
binuclear non-heme iron clusters. Chem. Rev. 96, 2625–2658 (1996).
8. Hohenberger, J., Ray, K. & Meyer, K. The biology and chemistry of high-valent
iron–oxo and iron–nitrido complexes. Nature Commun. 3, 720 (2012).
9. Nam, W. High-valent iron(IV)–oxo complexes of heme and non-heme ligands
in oxygenation reactions. Acc. Chem. Res. 40, 522–531 (2007).
10. Que, L. Jr The road to non-heme oxoferryls and beyond. Acc. Chem. Res.
40, 493–500 (2007).
Methods
Synthesis of Fe2(OH)0.6(dobdc) (2′) and Fe2(OH)2(dobdc) (2). An evacuated
Schlenk flask that contained fully desolvated Fe2(dobdc) (100 mg, 0.33 mmol)
was placed under an atmosphere of 30% N2O and 70% N2. The flask was immersed
in an oil bath, and the temperature was increased by 10 8C every 12 hours, from
25 8C up to 60 8C, to obtain Fe2(OH)2(dobdc) as a dark red–brown solid. When the
reaction was stopped after 12 hours at 35 8C, partially oxidized Fe2(OH)0.6(dobdc)
(as determined by Mo¨ssbauer spectroscopy) was obtained. Analytical: C8H4Fe2O8
calculated, C, 28.28, H, 1.19; found, C, 29.18, H, 1.16. Infrared (solid attenuated
total reflection (ATR)) spectroscopy: 3,679 (m), 1,532 (s), 1,450 (s), 1,411 (s),
1,361 (s), 1,261 (s), 1,154 (w), 1,129 (w), 1,077 (w), 909 (m), 889 (s), 818 (s),
807 (s), 667 (s), 630 (m), 594 (s), 507 (s).
11. Watton, S. P., Taylor, C. M., Kloster, G. M. & Bowman, S. C.
Coordination complexes in sol–gel silica materials. Prog. Inorg. Chem. 51,
333–420 (2002).
12. Leadbeater, N. E. & Marco, M. Preparation of polymer-supported ligands and
metal complexes for use in catalysis. Chem. Rev. 102, 3217–3274 (2002).
13. Panov, G. I. et al. Iron complexes in zeolites as a new model of methane
monooxygenase. React. Kinet. Catal. Lett. 61, 251–258 (1997).
14. Zecchina, A., Rivallan, M., Berlier, G., Lamberti, C. & Richhiardi, G. Structure
and nuclearity of active sites in Fe-zeolites: comparison with iron sites in
enzymes and homogeneous catalysts. Phys. Chem. Chem. Phys. 9,
3483–3499 (2007).
15. Yoon, J. W. et al. Controlled reducibility of a metal–organic framework with
coordinatively unsaturated sites for preferential gas sorption. Angew. Chem. Int.
Ed. 49, 5949–5952 (2010).
16. Ma, S., Yuan, D., Chang, J-S. & Zhou, H-C. Investigation of gas adsorption
performances and H2 affinities of porous metal–organic frameworks with
different entatic metal centers. Inorg. Chem. 48, 5398–5402 (2009).
17. Sumida, K. et al. Hydrogen storage and carbon dioxide capture in an iron-based
sodalite-type metal–organic framework (Fe–BTT) discovered via high-
throughput methods. Chem. Sci. 1, 184–191 (2010).
Synthesis of Fe0.1Mg1.9(dobdc) (3). In a 500-ml Schlenk flask, H4(dobdc) (1.8 g,
8.8 mmol), MgCl2 (1.5 g, 15 mmol) and FeCl2 (0.84 g, 6.6 mmol) were dissolved in a
mixture of 310 ml dimethylformamide (DMF) and 40 ml methanol. The reaction
was stirred vigorously at 120 8C for 16 hours. The precipitate was filtered and
stirred with 250 ml fresh DMF at 120 8C for three hours. Two more DMF washes at
120 8C were performed, after which the precipitate was filtered and soaked in
methanol at 60 8C. The methanol exchanges were repeated until no DMF
stretches were apparent in the infrared spectrum. The framework was desolvated
fully under dynamic vacuum (,15 mbar) at 210 8C for two days to afford
Fe0.1Mg1.9(dobdc) as a bright yellow–green solid (2.0 g, 8.2 mmol, 93% yield). The
iron-to-magnesium ratio was determined by inductively coupled plasma optical
emission spectrometry. Analytical: C8H2Fe0.1Mg1.9O6 calculated, C, 39.08, H,
0.82; found, C, 39.37, H, 0.43. Infrared (solid ATR) spectroscopy: 1,577 (s),
1,484 (m), 1,444 (s), 1,429 (s), 1,372 (s), 1,236 (s), 1,210 (s), 1,123 (m), 911 (m),
892 (s), 828 (s), 820 (s), 631 (s), 584 (s), 492 (s).
18. Bloch, E. D. et al. Selective binding of O2 over N2 in a redox-active metal–
organic framework with open iron(II) coordination sites. J. Am. Chem. Soc.
133, 14814–14822 (2011).
19. Bloch, E. D. et al. Hydrocarbon separations in a metal–organic framework
with open iron(II) coordination sites. Science 335, 1606–1610 (2012).
˚
20. Ma¨rcz, M., Johnsen, R. E., Dietzel, P. D. C. & Fjellvag, H. The iron
Reactivities of Fe2(dobdc) (1) and Fe0.1Mg1.9(dobdc) (3) with N2O and C2H6. In a
typical flow-through experiment, a mixture of gases (2 ml min21 N2O, 10 ml min21
C2H6 and 8 ml min21 Ar for a total flow 20 ml min21) was flowed over a packed bed
of MOF (50–100 mg) contained within a glass column. The column was heated to
75 8C for 24 hours, after which the products were extracted with CD3CN (3 × 1 ml)
and analysed by 1H NMR spectroscopy using 1,4-dichlorobenzene as an internal
standard. Although a cold bath maintained at –78 8C was installed downstream of
the glass reactor to collect condensable organic products, at the temperatures
tested all the products appeared to remain bound to the framework.
member of the CPO-27 coordination polymer series: synthesis,
characterization, and intriguing redox properties. Micropor. Mesopor. Mater.
157, 62–74 (2012).
21. Bhattacharjee, S. et al. Solvothermal synthesis of Fe-MOF-74 and its
catalytic properties in phenol hydroxylation. J. Nanosci. Nanotechnol. 10,
135–141 (2010).
22. Tolman, W. B. Binding and activation of N2O at transition-metal centers:
recent mechanistic insights. Angew. Chem. Int. Ed. 49, 1018–1024 (2010).
23. Piro, N. A., Lichterman, M. F., Harman, W. H. & Chang, C. J. A structurally
characterized nitrous oxide complex of vanadium. J. Am. Chem. Soc. 133,
2108–2111 (2011).
In a typical batch experiment, a Parr bomb was charged with N2O (1.5 bar)
and C2H6 (7.5 bar) and heated to 75 8C in a sand bath. After 24 hours, the
bomb was cooled and the products extracted with CD3CN.
24. Zhao, Y. & Truhlar, D. G. The M06 suite of density functionals for main group
thermochemistry, thermochemical kinetics, noncovalent interactions, excited
states, and transition elements: two new functionals and systematic testing of
four M06-class functionals and 12 other functionals. Theor. Chem. Acc.
120, 215–241 (2008).
Electronic structure calculations. The structures of 2 and 4 were optimized using
periodic boundary conditions and the PBE þ U exchange-correlation functional.
From each of these structures, we carved out a model cluster that contained three
iron centres along a single helical chain and six organic linkers. These clusters
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
5
© 2014 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1956
ARTICLES
25. Delabie, A., Vinckier, C., Flock, M. & Pierloot, K. Evaluating the activation
barriers for transition metal N2O reactions. J. Phys. Chem. A 105,
5479–5485 (2001).
44. Andersson, K., Malmqvist, P. A., Roos, B. O., Sadlej, A. J. & Wolinski, K. Second-
order perturbation theory with a CASSCF reference function. J. Phys. Chem.
94, 5483–5488 (1990).
26. Heyden, A., Peters, B. & Bell, A. T. Comprehensive DFT study of nitrous oxide
decomposition over Fe-ZSM-5. J. Phys. Chem. B. 109, 1857–1873 (2005).
27. Bottomley, F. & Brooks, W. V. F. Mode of bonding of dinitrogen oxide
(nitrous oxide) in (dinitrogenoxide)pentaammineruthenium. Inorg. Chem.
16, 501–502 (1977).
28. Pamplin, C. B., Ma, E. S. F., Safari, N., Rettig, S. J. & James, B. R. The nitrous
oxide complex, RuCl2(h1-N2O)(P–N)(PPh3) (P–N¼[o-(N,N-
dimethylamino)phenyl]diphenylphosphine); low temperature conversion of
N2O to N2 and O2. J. Am. Chem. Soc. 123, 8596–8597 (2001).
29. Paulat, F. et al. Spectroscopic properties and electronic structure of
pentammineruthenium(II) dinitrogen oxide and corresponding nitrosyl
complexes: binding mode of N2O and reactivity. Inorg. Chem. 43,
6979–6994 (2004).
30. Reed, A. E., Curtiss, L. A. & Weinhold, F. Intermolecular interactions from a
natural bond orbital, donor–acceptor viewpoint. Chem. Rev. 88, 899–926 (1998).
31. Macbeth, C. E. et al. O2 activation by nonheme iron complexes: a monomeric
Fe(III)–oxo complex derived from O2. Science 289, 938–941 (2000).
32. Dolphin, D. H., Sams, J. R., Tsin, T. B. & Wong, K. L. Mo¨ssbauer–Zeeman
spectra of some octaethylporphyrinato- and tetraphenylporphinatoiron(III)
complexes. J. Am. Chem. Soc. 100, 1711–1718 (1978).
33. Que, L. Jr & True, A. E. Dinuclear iron- and manganese-oxo sites in biology.
Prog. Inorg. Chem. 38, 97–200 (1990).
34. Soo, H. S., Komor, A. C., Iavarone, A. T. & Chang, C. J. A hydrogen-bond
facilitated cycle for oxygen reduction by an acid- and base-compatible iron
platform. Inorg. Chem. 48, 10024–10035 (2009).
35. Laarhoven, L. J. J., Mulder, P. & Wayner, D. D. M. Determination of bond
dissociation enthalpies in solution by photoacoustic calorimetry. Acc. Chem. Res.
32, 342–349 (1999).
36. Goldsmith, C. R., Jonas, R. T. & Stack, T. D. P. C–H bond activation by a ferric
methoxide complex: modeling the rate-determining step in the mechanism of
lipoxygenase. J. Am. Chem. Soc. 124, 83–96 (2002).
45. Shaik, S., Hirao, H. & Kumar, D. Reactivity of high-valent iron–oxo species
in enzymes and synthetic reagents: a tale of many states. Acc. Chem. Res.
40, 532–542 (2007).
46. England, J. et al. The crystal structure of a high-spin oxoiron(IV) complex
and characterization of its self-decay pathway. J. Am. Chem. Soc. 132,
8635–8644 (2010).
47. Lacy, D. C. et al. Formation, structure, and EPR detection of a high spin FeIV
–
oxo species derived from either an FeIII–oxo or FeIII–OH complex. J. Am. Chem.
Soc. 132, 12188–12190 (2010).
48. England, J. et al. A more reactive trigonal–bipyramidal high-spin oxoiron(IV)
complex with a cis-labile site. J. Am. Chem. Soc. 133, 11880–11883 (2011).
49. Bigi, J. P. et al. A high-spin iron(IV)–oxo complex supported by a trigonal
nonheme pyrrolide platform. J. Am. Chem. Soc. 134, 1536–1542 (2012).
50. Pestovsky, O. et al. Aqueous FeIV¼O: spectroscopic identification and
oxo-group exchange. Angew. Chem. Int. Ed. 44, 6871–6874 (2005).
Acknowledgements
Synthesis, basic characterization experiments and all of the theoretical work were
supported by the US Department of Energy, Office of Basic Energy Sciences, Division of
Chemical Sciences, Geosciences, and Biosciences under award DE-FG02-12ER16362.
Reactivity studies were supported by the Laboratory Directed Research and Development
Program of Lawrence Berkeley National Laboratory under US Department of Energy
Contract No. DE-AC02-05CH11231. Work at the Molecular Foundry, and XAS
experiments performed at the Advanced Light Source (BL 10.3.2), Berkeley, were supported
by the Office of Science, Office of Basic Energy Sciences, of the US Department of Energy
under Contract No. DE-AC02-05CH11231. X-ray diffraction experiments were performed
at the Advanced Photon Source at Argonne National Laboratory (17-BM-B). S.B., F.B. and
V.C. acknowledge financial support from the Ateneo Project 2011 ORTO11RRT5. We also
thank the National Science Foundation for providing graduate fellowship support (D.J.X.
and J.A.M.). In addition, we are grateful for the support of E.D.B. through a Gerald K.
Branch fellowship in chemistry, P.V. through a Phillips 66 Excellence Fellowship and
M.R.H. through the National Institute of Standards and Technology/National Research
Council Fellowship Program. We thank S. Chavan for help with the infrared spectroscopy
experiments and fruitful discussion.
37. Goldsmith, C. R. & Stack, T. D. P. Hydrogen atom abstraction by a mononuclear
ferric hydroxide complex: insights into the reactivity of lipoxygenase. Inorg.
Chem. 45, 6048–6055 (2006).
38. Starokon, E. V., Parfenov, M. V., Pirutko, L. V., Abornev, S. I. & Panov, G. I.
Room-temperature oxidation of methane by a-oxygen and extraction of
products from the FeZSM-5 surface. J. Phys. Chem. C 115, 2155–2161 (2011).
39. Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation
made simple. Phys. Rev. Lett. 77, 3865–3868 (1996).
40. Franchini, C. et al. Maximally localized Wannier functions in LaMnO3 within
PBEþU, hybrid functionals and partially self-consistent GW: an efficient
route to construct ab initio tight-binding parameters for eg perovskites. J. Phys.
Condens. Mater. 24, 235602/1-17 (2012).
Author contributions
D.J.X., E.D.B. and J.R.L. planned and executed the synthesis, characterization and reactivity
studies. J.A.M., W.L.Q., M.R.H. and C.M.B. analysed the powder neutron and X-ray
diffraction data. N.P. and P.V. performed the cluster DFT calculations. J.B. and K.L.
performed the periodic DFT calculations. A.L.D. performed the CASSCF/PT2 calculations.
D.G.T. and L.G. conceived and managed the computational efforts. F.B., V.C. and S.B.
carried out the in situ transmission Fourier transform infrared studies, and J.Y. supervised
EXAFS analysis. All authors participated in the preparation of the manuscript.
41. Verma, P., Xu, X. & Truhlar, D. G. Adsorption on Fe-MOF-74 for C1–C3
hydrocarbon separation. J. Phys. Chem. C 117, 12648–12660 (2013).
42. Maurice, R. et al. Single-ion magnetic anisotropy and isotropic magnetic
couplings in the metal–organic framework Fe2(dobdc). Inorg. Chem. 52,
9379–9389 (2013).
Additional information
Supplementary information is available in the online version of the paper. Reprints and
permissions information is available online at www.nature.com/reprints. Correspondence and
requests for materials should be addressed to J.R.L.
43. Andersson, K., Malmqvist, P-Å. & Roos, B. O. Second order perturbation theory
with a complete active space self-consistent field reference function. J. Chem.
Phys. 96, 1218–1226 (1992).
Competing financial interests
The authors declare no competing financial interests.
6
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
© 2014 Macmillan Publishers Limited. All rights reserved.