Metal-dioxidoterephthalate MOFs of the MOF-74 type: Microporous basic catalysts with well-defined active sites

Article abstract of DOI:10.1016/j.jcat.2014.06.006

The hybrid frameworks M₂dobdc (dobdc^4- = 2,5-dioxidoterephthalate, M²⁺ = Mg²⁺, Co²⁺, Ni²⁺, Cu²⁺ and Zn²⁺), commonly known as CPO-27 or MOF-74, are shown to be active catalysts in base-catalyzed reactions such as Knoevenagel condensations or Michael additions. Rather than utilizing N-functionalized linkers as a source of basicity, the intrinsic basicity of these materials arises from the presence of the phenolate oxygen atoms coordinated to the metal ions. The overall activity is due to a complex interplay of the basic properties of these structural phenolates and the reactant binding characteristics of the coordinatively unsaturated sites. The nature of the active site and the order of activity between the different M ₂dobdc materials were rationalized via computational efforts; the most active material, both in theory and in experiment, is the Ni-containing variant. The basicity of Ni₂dobdc was experimentally proven by chemisorption of pyrrole and observation by IR spectroscopy.

Full text of DOI:10.1016/j.jcat.2014.06.006

Journal of Catalysis 317 (2014) 1–10
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Metal-dioxidoterephthalate MOFs of the MOF-74 type: Microporous
basic catalysts with well-defined active sites
Pieterjan Valvekens a, Matthias Vandichel b, Michel Waroquier b, Veronique Van Speybroeck b,
,
⇑
Dirk De Vos a,
⇑
a Centre for Surface Chemistry and Catalysis, KU Leuven, Kasteelpark Arenberg 23, Box 2461, 3001 Leuven, Belgium
b Center for Molecular Modeling, Universiteit Gent, Technologiepark 903, 9052 Zwijnaarde, Belgium
a r t i c l e i n f o
a b s t r a c t
The hybrid frameworks M2dobdc (dobdc4À = 2,5-dioxidoterephthalate, M2+ = Mg2+, Co2+, Ni2+, Cu2+ and
Zn2+), commonly known as CPO-27 or MOF-74, are shown to be active catalysts in base-catalyzed reac-
tions such as Knoevenagel condensations or Michael additions. Rather than utilizing N-functionalized
linkers as a source of basicity, the intrinsic basicity of these materials arises from the presence of the phe-
nolate oxygen atoms coordinated to the metal ions. The overall activity is due to a complex interplay of
the basic properties of these structural phenolates and the reactant binding characteristics of the coord-
inatively unsaturated sites. The nature of the active site and the order of activity between the different
M2dobdc materials were rationalized via computational efforts; the most active material, both in theory
and in experiment, is the Ni-containing variant. The basicity of Ni2dobdc was experimentally proven by
chemisorption of pyrrole and observation by IR spectroscopy.
Article history:
Received 11 February 2014
Revised 2 June 2014
Accepted 4 June 2014
Keywords:
Metal–organic framework
CPO-27/MOF-74
Basic catalysis
Knoevenagel condensation
Michael addition
Phenolate oxygen
Ó 2014 Elsevier Inc. All rights reserved.
1. Introduction
ZIF-8 and ZIF-9 [9–11], in which the pore windows are too narrow
to allow diffusion of most reactants and/or products, or a Ba-3,5-
Metal–organic frameworks (MOFs) have attracted much inter-
est due to their large compositional and structural diversity. A
myriad of applications have been investigated so far, in domains
such as gas storage, separations, and catalysis [1–3]. The explora-
tion of the catalytic potential of MOFs, however, has mainly
focused on acid or redox catalysis. Reports on the catalytic activity
of MOFs in base-catalyzed reactions are much more scarce, and the
reported materials can be divided into two groups. A first group
consists of porous materials containing N-functionalized ligands,
e.g. aminoterephthalic acid [4–7]. In these materials, the N-func-
tionality acts as the basic catalytic site. The relatively weak basicity
of these sites, however, with electron withdrawing carboxylic
groups on the same aromatic ring, limits the reaction scope of
these catalysts, and some of the early reports, e.g. on amino-substi-
tuted MOF-5, have been reinterpreted in terms of structural defects
of water-labile Zn frameworks [8]. The second group of materials
comprises structures lacking the porosity requirements for intra-
porous catalysis, featuring basic sites mainly at lattice terminating
surface sites. Examples of such materials are the Zn-imidazolates
pyrazoledicarboxylate MOF having a limited microporous volume
(SBET = 56 m2/g) [12]. If catalysis is confined to the outer surface,
the catalytic material is used only for a very limited fraction, and
the precise nature of the sites at the surface is not always well
defined. The group of Koner also synthesized and used various
MOFs combining Mg or Ba and pyrazole-3,5-dicarboxylic acid or
pyridine-2,5-dicarboxylic acid [13–15]. Although these materials
were catalytically active in aldol condensation reactions, the pres-
ence of stoichiometric amounts of the base triethylamine was
required to obtain sufficient reaction rates.
To the best of our knowledge, there are no studies of porous
MOFs with an intrinsic framework basicity that is useful in classi-
cal base-catalyzed reactions such as Knoevenagel condensations or
conjugate additions. We here report that surprisingly, M2dobdc-
type MOFs are active catalysts for such reactions, with intrinsically
basic properties. M2dobdc represents a series of isostructural com-
pounds, denominated CPO-27-M by Dietzel et al. [16–20], or MOF-
74 for the zinc compound by Rosi et al. [21] (CPO: Coordination
Polymer of Oslo; M = Ni, Co, Zn, Mg, Mn). The structure consists
of metal oxide chains connected by the 2,5-dioxidoterephthalate
linker (dobdc4À) forming a honeycomb-like structure with large,
one-dimensional hexagonal pores of 1.1–1.2 nm in diameter
(Fig. 1 (top)). In Fig. 1 (bottom), the potentially available base sites
⇑
Corresponding authors.
E-mail addresses: veronique.vanspeybroeck@ugent.be (V. Van Speybroeck), dirk.
devos@biw.kuleuven.be (D. De Vos).
http://dx.doi.org/10.1016/j.jcat.2014.06.006
0021-9517/Ó 2014 Elsevier Inc. All rights reserved.

2
P. Valvekens et al. / Journal of Catalysis 317 (2014) 1–10
chemicals were of the highest grade available and were used with-
out further purification.
2.2. Catalyst preparation
Mg2dobdc, Co2dobdc, Ni2dobdc, Cu2dobdc, and Zn2dobdc were
synthesized according to literature procedures [26,35–37]. The
synthesis solvent was removed by washing the material seven
times with methanol over a period of four days. Prior to catalytic
testing, the materials were activated under vacuum, <10À5 bar, at
elevated temperature according to literature procedures, i.e.
130 °C, 200 °C, 240°, 250 °C, and 270 °C for Cu2-, Mg2-, Ni2-, Co2-,
and Zn2dobdc respectively. After activation, the materials were
transferred under inert conditions to the reaction vessel.
2.3. Catalyst characterization
The crystallinity of the synthesized materials was confirmed via
powder X-ray diffraction (PXRD). Reflection patterns were
recorded on a STOE STADI MP in Bragg–Brentano mode (2h–h
geometry; Cu Ka1, 1.54060 Å) using a linear position-sensitive
detector. N2 sorption measurements were taken on a Micromeri-
tics 3Flex surface analyzer at 77 K. Prior to measurements, the
samples were activated under vacuum at the temperatures
reported in the literature (vide supra). Fourier transform infrared
spectroscopy measurements (FTIR) were taken on a Nicolet 6700
spectrometer. Thin self-supporting wafers ( 10 mg cmÀ2) were
prepared and introduced in an in situ cell. The cell was brought
under vacuum (p < 10À4 Pa) at 240 °C, and IR spectra of the evacu-
ated sample were recorded at different temperatures during the
subsequent cooling period. Pyrrole was adsorbed at room temper-
ature from the vapor and the weakly physisorbed pyrrole was
evacuated at 50 °C, after which the spectrum was recorded and
compared with the spectrum of the original sample at 50 °C.
Fig. 1. (top) Pores in the M2dobdc MOF; a sphere with a diameter of 1.2 nm fits
inside the pores (brown = carbon; orange = metal; red = oxygen); (bottom) repre-
sentation of the potential base sites and open metal sites in the framework. (For
interpretation of the references to color in this figure legend, the reader is referred
to the web version of this article.)
in the framework are displayed, together with the well-known
coordinatively unsaturated sites [22].
The high density of exposed metal ions in these materials has
sparked much interest in the field of adsorption and the adsorptive
separation of compounds such as CO, CO2, CH4, ethane, acetylene,
propane, propylene, or H2S [20,23–31]. The existence of a family
of isostructural materials has also allowed to study the specific role
of the metal ion in these adsorption processes [20,24,29,32].
Reports of catalysis using M2dobdc are very limited hitherto; they
focus on the role of the metal ion as a Lewis acid center or in accel-
erating autoxidation reactions [22,33,34]. The present study
focuses on the use of the Mg2+, Co2+, Ni2+, Cu2+, and Zn2+ variants
as catalysts in Knoevenagel condensation and Michael conjugate
addition reactions and on understanding the activation of the reac-
tants on the phenolate–metal active sites. The successful applica-
tion of these MOFs for these standard base-catalyzed reactions
opens a new window for catalysis research using the intrinsic basi-
city of MOFs.
2.4. Catalyst evaluation
In a typical reaction, 1 mmol of each reactant was combined
with 2 ml of solvent, toluene for the Knoevenagel condensations
or acetonitrile in the Michael addition. The mixture was added to
50 mg of activated catalyst. The reaction mixture was heated to
70 °C, and aliquots of sample were removed for analysis at set
intervals. The identity of the reaction products was verified by
GC–MS (Agilent 6890 gas chromatograph, equipped with a HP-
5MS column, coupled to a 5973 MSD mass spectrometer), and
the product yields were determined via GC-analysis. The substrate
scope was evaluated in reactions in which 1 mmol of each reactant
was combined with 2 ml of toluene. This mixture was added to
50 mg of appropriately activated Ni2dobdc and stirred at 110 °C.
2.5. Theoretical modeling
2. Materials and methods
The theoretical part of this study focuses on the key steps initi-
ating the Knoevenagel condensation reactions, in an attempt to
understand the order of activity as observed experimentally for
the different M2dobdc materials. In order to get a first vision on
the basicity of the material and on the characteristics of the open
metal sites, a preliminary study has been performed by evaluating
the proton affinities of the pristine nanoporous materials and their
adsorption properties with respect to some small probe molecules.
Periodic DFT-D calculations were carried out on all M2dobdc
structures (M = Co, Ni, Cu, Mg, Zn) with the Vienna Ab Initio Simu-
lation Package (VASP 5.2.12) [38–41]. For each metal, a 2 Â 1 Â 1
unit cell was optimized with a plane wave kinetic energy cutoff
of 550 eV, employing the PBE exchange–correlation functional
[42,43] with D2-dispersion corrections of Grimme and coworkers
2.1. Materials
Copper(II) nitrate trihydrate, malononitrile, and 2-cyclohexene-
1-one were purchased from Fluka; N,N-dimethylformamide, N-
methylpyrrolidone, zinc nitrate hexahydrate, benzaldehyde, and
2-cyclopentene-1-one from Acros; methanol, toluene, and acetoni-
trile from VWR; magnesium acetate tetrahydrate, ethyl cyanoace-
tate, ethyl acetoacetate, and acetophenone from Aldrich;
tetrahydrofuran and methyl vinyl ketone from Sigma-Aldrich;
cobalt nitrate hexahydrate from Alfa Aesar; nickel acetate tetrahy-
drate, diethyl malonate, cyclopentanone, and cyclohexanone from
Janssen Chimica and 2,5-dihydroxyterephthalic acid from TCI. All

P. Valvekens et al. / Journal of Catalysis 317 (2014) 1–10
3
[44], followed by a single point energy computation at the PBE-D3
level of theory. The projector augmented wave approximation
(PAW) [45] was used. Brillouin zone sampling was restricted to
the U-point. Gaussian smearing [39] was applied to improve con-
vergence: 0.05 eV for cell optimizations and energy calculations.
For the cell optimizations, the convergence criterion for the elec-
tronic self-consistent field (SCF) problem was set to 10À6 eV, while
the atomic forces were converged below 0.01 eV/Å (structures are
reported in Supporting information). For the further applications,
the cell parameters were kept fixed, the kinetic energy cutoff
was lowered to 400 eV, and the atomic force criterion was set on
0.03 eV/Å (for the adsorption and deprotonation reactions).
Furthermore, we used spin polarization throughout all calculations
applying the most stable spin state for the metal in gas phase for
every metal atom in the crystal. The reported adsorption and
deprotonation energies do not include zero-point vibrational ener-
gies or thermal corrections.
Scheme 1.
In a first exploratory survey of the nature of the active site,
adsorption energies of some basic probe molecules on the coordin-
atively unsaturated sites (cus) were calculated. Therefore, we used
the methodology as outlined above for the periodic structure cal-
culations. Another bulk property, which is indicative for the base
strength, is the proton affinity of the material (PA). Periodic calcu-
lations fail in reproducing accurate proton affinities or deprotona-
tion energies of periodic materials, due to the charge of the proton
induced in the unit cell when adding or removing a proton from
the framework; therefore, cluster calculations were preferred.
The cluster models were cut from the optimized periodic struc-
tures (see Fig. S.2), which in turn are further optimized with the
Gaussian09 package [46], using the B3LYP hybrid functional
[47,48] including D3 Grimme corrections for the van der Waals
interaction [44]. The double-zeta Pople basis set 6–31g(d) was
used for all atoms. By fixing the outer carboxyl oxygen atoms, we
maintained the rigidity of the cluster model as in the crystal. The
manipulation of the periodic structure files and the cluster models
was performed using ZEOBUILDER, an in house developed software
tool for building complex molecular structures [49]. Partial Hessian
vibrational analysis has been post-processed using the program
TAMKIN [50].
Knoevenagel condensation, we found experimentally an overall
order of activity of Ni > Cu > Mg ꢀ Zn > Co. Replacing MN by the
more weakly acidic ECA as donor, the conversion occurs signifi-
cantly more slowly, and after 2 h, no clear discrimination between
the yields over the various catalysts can be extracted. Very roughly,
we found following order of activity: Ni ꢀ Mg ꢀ Zn > Co ꢀ Cu (see
Table 1). The lower reactivity of ECA vs. MN was also observed in
other catalytic studies [54,56,57].
To investigate whether similar activity trends can be observed
in other base-catalyzed reactions, an additional Michael addition
of ethyl cyanoacetate to methyl vinyl ketone (Scheme 1) was
attempted with the same series of catalysts (Table 1). Despite the
relatively weak acidity of ethyl cyanoacetate, high conversions
are observed, with again the Ni variant as the most active member
of the catalyst series.
3.2. Detailed study of reactions with the Ni2dobdc catalyst
In view of its high activity, the Ni2dobdc catalyst was employed
for studying the reactivity of different starting compounds, in an
attempt to obtain more insight in how the reactants are activated.
To this end, both the Knoevenagel condensation and the Michael
addition reactions were performed in toluene at 110 °C. In the
Knoevenagel condensations with benzaldehyde (Table 2), active
methylene compounds were tested with pKa values ranging
between 11.1 and 16.4. When varying the donor molecules, their
pKa is clearly decisive for the conversion rates (entries 1–4); the
highest conversion was observed for the most acidic compound,
MN, while both ethyl acetoacetate and diethyl malonate show
much lower conversions. Variation in the acceptor molecules
shows that the presence of a methyl substituent on the carbonyl
group, like in acetophenone, decreases the reactivity of this mole-
cule, which is probably due to steric hindrance (entry 5). Reactions
with cyclopentanone and cyclohexanone (entries 6 and 7) demon-
strate the possibility to use also cyclic aliphatic ketones as reac-
tants in Knoevenagel condensation reactions catalyzed by
Ni2dobdc.
While the reactivity of the donor molecules in the Knoevenagel
condensation is dominated by the pKa of these molecules, the
donor reactivity order in the Michael addition reactions over Ni2-
dobdc is more complex (Table 3). At first glance, there appears to
be a reversal in the expected, pKa-based reactivity order: in these
Michael reactions, the lowest conversions were observed for the
most acidic donor molecule, malononitrile (entry 1), whereas the
highest conversion, even after short time, was observed for the
not so acidic ethyl acetoacetate (entry 4). The conversion of diethyl
malonate, the least acidic compound (entry 5), is, however, again
3. Results and discussion
3.1. Activity of M2dobdc materials with various structural cations
All materials were synthesized and pretreated according to lit-
erature procedures, and PXRD patterns and BET surface areas of the
synthesized materials can be found in Fig. S.1 and Table S.1.
Knoevenagel condensations are widely studied for MOFs, and some
excellent catalysts have already been reported. These range from
purely basic MOFs with N-functionalized groups [4,9,51], to MOFs
featuring Lewis acid active sites such as Cu3(BTC)2, Fe3(BTC)2 or
Ni4(MTB)2 (H3BTC = trimesic acid, H4MTB = methanetetrabenzoic
acid) [52–55]. The catalytic activity of M2dobdc in some typically
base-catalyzed reactions was therefore first evaluated via the
Knoevenagel condensation of malononitrile (MN) and benzalde-
hyde (BA) (as shown in Scheme 1) and of ethyl cyanoacetate
(ECA) and benzaldehyde. In order to take into account the lower
atomic weight of magnesium compared to the transition metal
ions tested, the amount of Mg2dobdc used in these tests was
adapted accordingly. The catalytic data (Table 1) show a distinct
influence of the metal ion on the performance of the catalyst, with
a clear preference for Ni2dobdc (entry 3) as the most active mate-
rial in the reaction of MN and BA. These trends are based on initial
rates (X2h), which should be only minimally affected by the
possible adsorption of the reaction products. For the MN + BA

4
P. Valvekens et al. / Journal of Catalysis 317 (2014) 1–10
Table 1
Conversion (X) measured after 2 and 24 h of reaction for the different M2dobdc materials in Knoevenagel condensation and Michael addition reactions.a
M
Knoevenagel condensation
MN + BA
Knoevenagel condensation
ECA + BA
Michael addition
ECA + MVK
X2h
X24h
X2h
X24h
X2h
X24h
1
2
3
4
5
Mg
Co
Ni
Cu
Zn
17
4
69
39
13
73
32
99
69
77
3
0
3
0
3
14
9
9
9
14
6
9
26
13
19
35
42
66
38
44
a
Reaction conditions: 1 mmol of each reactant; 2 ml solvent, 343 K, 50 mg of M2dobdc material (M = Co, Ni, Cu, Zn) or 37 mg (M = Mg); MN = malononitrile, BA = benz-
aldehyde; ECA = ethyl cyanoacetate; MVK = methyl vinyl ketone.
Table 2
mixture (Fig. 2). Clearly, in such complex competitive reactions,
many effects could play a role, e.g. not only differential occupation
of the active sites by reactants, intermediates or products, but also
different intraporous concentrations when the sorption equilib-
rium is attained. Such complexity is beyond the scope of the theo-
retical work later in this paper, which will mainly address the
activation of acidic donor molecules on the M2dobdc-active sites.
Finally, the reaction between malononitrile and 2-cyclopenten-
1-one or 2-cyclohexen-1-one was studied (Table 4). According to
the literature, the nature of the catalyst strongly influences
whether a Knoevenagel or a Michael addition reaction takes place.
Whereas Lewis acids such as RuCl3 or Ru-exchanged hydroxyapa-
tite are reported to selectively catalyze the Knoevenagel condensa-
tion [58], bases such as 1,1,3,3-tetramethylguanidine [59] or
t-BuOK [60] are reported to catalyze the Michael addition reaction
of these compounds. Use of Ni2dobdc as a catalyst resulted in the
formation of both the Michael addition and the Knoevenagel con-
densation products (Table 4). In the reaction with 2-cyclopenten-
1-one, the Michael addition product is the primary product; with
2-cyclohexen-1-one, the main product is the Knoevenagel product.
The presence of both the Michael and Knoevenagel reaction prod-
ucts provides evidence for the simultaneous action of base and
open metal sites on the catalyst.
Evaluation of the substrate scope for Knoevenagel condensation reactions with
Ni2dobdc.a
Donor (pKa in DMSO)
Acceptor
t (h)
Xacceptor (%)
1
2
3
4
5
6
7
Malononitrile
(11.1)
(13.1)
(14.3)
(16.4)
(11.1)
(11.1)
(11.1)
Benzaldehyde
Benzaldehyde
Benzaldehyde
Benzaldehyde
Acetophenone
Cyclopentanone
Cyclohexanone
24
24
24
24
24
24
24
100b
25
3
<1
13
52
87
Ethyl cyanoacetate
Ethyl acetoacetate
Diethyl malonate
Malononitrile
Malononitrile
Malononitrile
a
Standard reaction conditions: 1 mmol of each reactant, 50 mg of catalyst, 2 ml of
toluene, 110 °C. Xacceptor = conversion of the acceptor molecule;
b
Reaction at 70 °C.
Table 3
Evaluation of the substrate scope for Michael addition reactions with Ni2dobdc, using
methyl vinyl ketone as the acceptor.a
Donor (pKa in DMSO)
t (h)
Xacceptor (%)
1
2
3
4
5
Malononitrile
(11.1)
(13.1)
(13.3)
(14.3)
(16.4)
2 (24)
2 (24)
2 (24)
2 (24)
2 (24)
18 (55)
32 (75)
58 (99)
93 (99)
66 (86)
Ethyl cyanoacetate
2,4-Pentanedione
Ethyl acetoacetate
Diethyl malonate
The heterogeneity of the Ni2dobdc catalyst was evaluated via a
hot filtration test on a Knoevenagel reaction (Fig. 3, top). Removal
of the catalyst after both 1 and 2 h of reaction confirms that catal-
ysis is indeed fully heterogeneous. Furthermore, PXRD measure-
ments before and after reaction confirm the stability of the
a
Standard reaction conditions: 1 mmol of donor, 1 mmol of methyl vinyl ketone as
acceptor, 50 mg of catalyst, 2 ml of toluene, 110 °C.
slightly lower than that of ethyl acetoacetate. In order to explain
such reactivity trends, we hypothesize that not only the donor,
but also the acceptor could be activated during the Michael reac-
tion. In such a mechanism, a basic site abstracts a proton from
the donor molecule, whereas the
a,b-unsaturated carbonyl com-
pound could be activated via coordination on an open metal site.
However, if the anion is more easily formed from the donor, as
for the more acidic donor molecules, or if there is an excess of coor-
dinating nitrile functional groups, like in malononitrile (entry 1),
the coordinatively unsaturated sites may be fully occupied by
either carbanions derived from the donor, or the nitrile groups.
Such a coordination may hamper the activation of the acceptor
molecule, slowing down the overall reaction. This could explain
why malononitrile containing two nitrile functions would give
the lowest rate, closely followed by ethyl cyanoacetate. Such
site-blocking effects are less expected for donor molecules lacking
nitrile functions, such as 2,4-pentanedione, ethyl acetoacetate, or
diethyl malonate (entries 3–5). In order to investigate the effect
of the presence of nitriles on the conversion, an additional compet-
itive Michael reaction was carried out using both malononitrile
and ethyl cyanoacetate as donor molecules. This reaction shows
that, whereas the conversion of malononitrile is relatively uninhib-
ited by ethyl cyanoacetate, the conversion of ethyl cyanoacetate is
strongly inhibited by the presence of malononitrile in the reaction
Fig. 2. Conversion of ethyl cyanoacetate and malononitrile in the Michael addition
reaction with methyl vinyl ketone over Ni2dobdc after 24 h, in individual
experiments (with one donor) and in a competitive experiment (with two donors).
Reaction conditions: 0.5 mmol donor, 1 mmol methyl vinyl ketone, 50 mg of catalyst,
2 ml of toluene, 110 °C. In the competitive experiment, 0.5 mmol of each donor
molecule and 1 mmol of methyl vinyl ketone was used.

P. Valvekens et al. / Journal of Catalysis 317 (2014) 1–10
5
Table 4
Reactions of 2-cycloalken-1-one acceptors with malononitrile donor using Ni2dobdc catalysts.a
Acceptor
t (h)
Yield Knoevenagel product (%)
Yield Michael product (%)
X (%)b
1
2
2-Cyclopenten-1-one
2-Cyclohexen-1-one
24
24
5
27
23
7
27
35
a
Reaction conditions: 2-cycloalken-1-one (1 mmol), malononitrile (1 mmol), 50 mg Ni2dobdc catalyst, toluene (2 ml), 110 °C.
Conversion of the acceptor.
b
catalyst structure under reaction conditions (Fig. 3, middle) and a
recycle test confirms that the catalyst can be reused over several
runs. Scale-up of a catalytic reaction was successfully performed;
using 10 times larger quantities of the reactants and the catalyst
than in standard conditions, 1.47 g of
(95% yield) was obtained after 24 h.
a-cyanocinnamonitrile
3.3. Spectroscopic observation of the chemisorption of pyrrole
Next, the chemisorption of pyrrole, as an acidic probe molecule,
was monitored using FTIR spectroscopy in order to experimentally
prove that the M2dobdc materials possess base sites next to the
well-known open metal sites [61–63]. Two distinct signals can be
observed in the FTIR difference spectrum (Fig. 4). A first absorption
band at 3405 cmÀ1 typically corresponds to the stretching vibra-
tion of the NH group interacting via an NH–p complex with the
ring of another pyrrole molecule [64,65]. The second and most
intense absorption band in the diagnostic region of pyrrole can
be found for Ni2dobdc at 3253 cmÀ1. Although the exact position
of this absorption band can be influenced by different interactions,
such as the interaction of the aromatic ring of pyrrole with the
open metal site, the strong bathochromic shift of this band com-
pared to that of free pyrrole is generally attributed to pyrrole spe-
cies interacting with a H-bond acceptor stronger than pyrrole, for
instance a basic site on the surface [63–65]. The shift of the NAH
stretching vibration frequency in comparison with that of free pyr-
role, at 3500 cmÀ1, is a measure for the basic strength of the
adsorption site. A shift of 250 cmÀ1 is comparable to the shift as
observed for the interaction of pyrrole with alumina [65] or pyri-
dine [64]. Similar results were obtained for Mg2-, Co2- and Zn2-
dobdc; these results are summarized in Table S.2 (Supporting
information).
Using dynamic simulations in periodic code (NVT-CP2K-PBE-
D3, time step 0.5 fs, 373 K, simulation time = 14.69 ps), no distinct
adsorption positions for pyrrole could be found. A logical adsorp-
tion mode is one in which the pyrrole has its NAH oriented toward
the phenolate oxygen. For such an adsorption mode, calculation of
the NAH stretching frequency in the static mode using periodic
code (VASP-PBE-D2) showed that in comparison with free pyrrole,
the adsorbed pyrrole experiences a 244 cmÀ1 downshift of the
stretching frequency, which is in remarkably good agreement with
the experimentally observed value. This supports the vision of the
phenolate oxygen atom acting as a base toward the pyrrole probe.
3.4. Theoretical rationalization of the catalytic activity
The catalytic activity of M2dobdc as a base can be understood by
looking at the structure of this MOF. As the linker molecule, 2,5-
dihydroxyterephthalic acid, is fully deprotonated in the framework,
carboxylate oxygen atoms as well as phenolate oxygen atoms coor-
dinate the metal center (Fig. 1 (bottom)). While the carboxylate oxy-
gen atoms are only weakly basic, a phenolate oxygen, which is a
conjugate base to the weakly acid phenol, is expected to be much
more basic. Logically, these phenolate oxygen atoms should act as
the basic site in this material, deprotonating the acidic reactant,
e.g. malononitrile or ethyl cyanoacetate, thus activating this com-
pound for further reaction. The open coordination site on the metal
Fig. 3. Hot filtration test for Ni2dobdc in the Knoevenagel condensation of
malononitrile with benzaldehyde (top), PXRD diffraction patterns before and after
reaction (middle) and product yield in consecutive runs with the same catalyst
(bottom).

6
P. Valvekens et al. / Journal of Catalysis 317 (2014) 1–10
Table 5
Proton affinities of M2dobdc at 0 K, for the phenolate oxygen atom (PA1, kJ/mol), and
for the carboxylate oxygen atom (PA2, kJ/mol). Also given are the computed
adsorption energies of the basic probe molecules acetonitrile and carbon monoxide
(D
Eads, kJ/mol).
M
PA1
PA2
D
Eads (CH3CN)
DEads (CO)
1
2
3
4
5
Mg
Co
Ni
Cu
Zn
930
947
946
923
944
913
918
757
905
917
À91
À83
À91
À47
À79
À45
À92
À79
À27
À39
as the coordination of the phenolate to the metal ion influences
the electron density of the oxygen atom. As a first indication for
the basicity of the material, the proton affinity (PA) was calculated
in an attempt to explain the experimentally observed conversion
order in both the Knoevenagel condensation and Michael addition
reactions. PAs were calculated for both the phenolate (PA1) as well
as for the carboxylate oxygen atom (PA2) (Fig. 1, bottom) and are
tabulated in Table 5. As expected, the protonation is favored on
the phenolate oxygen atoms, as PA1 is systematically larger than
PA2 for all metals. Hence, deprotonation of the donor molecules
(MN, ECA) will preferably take place on the phenolate oxygen
atoms. The Ni, Co, and Zn materials show the largest proton affin-
ity, while Cu behaves somewhat differently in the sense that the
basicity of the phenolate oxygen is dramatically reduced, coming
closer to that of the carboxylate oxygen atoms. The performance
of the Cu material in the experiments also strongly varies depend-
ing on the choice of the acidic donor. While the high proton affinity
of the phenolate in e.g. Ni2dobdc is consistent with the high
observed activity for this material, it is clearly impossible to ratio-
nalize all reactivity trends based on the crude PA values alone. In
the following subsection, the deprotonation will be examined in
the reaction itself, with explicit consideration of the nature of the
acidic donor molecule and of the presence of one or more adjacent
open metal sites.
Fig. 4. FTIR difference spectrum of pyrrole chemisorbed on Ni2dobdc.
ion adjacent to the phenolate oxygen atom can act as a docking site
for the deprotonated reactant molecule. The second reactant in the
Knoevenagel condensation, e.g. benzaldehyde, can be adsorbed on
the protonated phenolate oxygen, after which coupling of the reac-
tants takes place as shown in the reaction mechanism displayed in
Scheme 2. Such an interplay of metal sites and phenolate oxygen
atoms in the formation of acid–base reactive centers in catalysts is
similar to what is happening for homogeneous alkaline earth metal
binaphtholate catalysts described by e.g. Kobayashi and Yamashita
[66] or Hatano et al. [67], for barium and magnesium ions, respec-
tively. In the following subsections, molecular modeling will be used
for (1) the characterization of the basic sites and the open metal sites
in the M2dobdc materials and (2) studying the initial steps in the
reaction mechanism for the Knoevenagel condensation (experimen-
tal data in Table 1).
Regarding the characteristics of the open metal site, the adsorp-
tion energy of some small Lewis basic probe molecules (CO, aceto-
nitrile) on the framework may be used as an indicative factor. In
3.4.1. The phenolate–metal active site
A priori, one would expect that the basicity of the M2dobdc
materials would be affected by the nature of the metal ion M2+
,
Scheme 2. The proposed reaction mechanism for the Knoevenagel condensation of malononitrile and benzaldehyde in M2dobdc.

P. Valvekens et al. / Journal of Catalysis 317 (2014) 1–10
7
Fig. 5. Schematic representation of adsorption and subsequent deprotonation reaction of malononitrile (MN), and corresponding energy profiles for different M2dobdc
materials. MN can be adsorbed on one (b) or two adjacent metal sites (c). Note that the deprotonation of MN, adsorbed at two metal sites, can only occur if one nitrile–metal
coordination is released (d).
order to quantify the possible roles of cus sites in the reactions
under scrutiny, the adsorption energies of such probe molecules
on the cus were studied. Heats for adsorption of CO in M2dobdc
materials (M = Ni, Mg, Zn) have been recently calculated by Val-
enzano et al. [68] using DFT and hybrid MP2/DFT methods. Of
these metals, Ni is the one with the largest adsorption energy for
CO. Values for Mg and Zn are closer together, but all level of theory
methods applied by Valenzano predict the order: Ni > Mg > Zn. For
completeness, we also performed adsorption calculations for a lar-
ger set of metals. The results are reported in Table 5 and the fol-
lowing trend was found: Co > Ni > Mg > Zn > Cu, in agreement
with Valenzano et al. Reported values for the adsorption of CO
on Ni2dobdc are remarkably larger than those predicted by Valenz-
ano et al. [68]. The high affinity of Ni2dobdc and Co2dobdc for CO
can be rationalized by considering that the cus in these materials
adsorption energies is the result for Cu, manifesting itself as an
outlier in generating low predicted values for the adsorption of
the two probe molecules.
Summarizing, the ranking of the metals with respect to the
energies of adsorption of Lewis bases seems different for each
probe molecule investigated; however, the Ni material, which we
selected as a preferred catalyst, invariably shows one of the stron-
gest interactions, irrespective of the nature of the Lewis base. On
the other hand, when considered itself as a base, the Ni2dobdc
material also displays a considerable PA. Thus, the simulations do
not indicate that there would be an inverse relationship between
the adsorption energies of probe molecules on the open metal sites
and the basicity of the adjacent framework phenolate. This sug-
gests that one and the same material could contain both active
base and reactive open metal sites.
not only can accept an electron pair, but also can provide
donation to the CO [28,61,69–71].
p-back
3.4.2. Mechanism of basic reactions on the active site and
rationalization of activity
Also incorporated in Table 5 are the adsorption energies of ace-
tonitrile on the various M2dobdc materials, as we expect a similar
adsorption sequence for the donors with nitrile groups, such as MN
or ECA. Inspection of Table 5 learns that the correlation between
the two series of adsorption energies, for CO and acetonitrile,
respectively, is rather poor. In addition, the adsorption energies
vary over a broad range of values. The most striking case of a dif-
ferent adsorption pattern is offered by the Mg material, with a dif-
ference of almost 50 kJ/mol in adsorption energy between
acetonitrile and CO. One common feature in the two series of
We now focus on the reaction steps initiating the Knoevenagel
condensation, namely the adsorption of the donor molecule on the
metal site, and the subsequent deprotonation on the adjacent phe-
nolate as schematically shown in Figs. 5 and 6. We consider two
donors, MN and the less acidic ECA, and take into account different
adsorption positions of the donors. MN has two nitrile groups of
which just one or two could be adsorbed on neighboring metal
sites (Fig. 5b and c). The energies associated with the different
stages are also plotted for the different metals.

8
P. Valvekens et al. / Journal of Catalysis 317 (2014) 1–10
Fig. 6. Schematic representation of adsorption and subsequent deprotonation reactions of ethyl cyanoacetate (ECA) and corresponding energy profile for different M2dobdc
materials. ECA can be adsorbed with the nitrile group (b), but also with the carboxyl oxygen (d). Note that the end product obtained after deprotonation of the adsorbed
complex with the carboxyl oxygen coordinated with the metal (e), is not susceptible to further reaction with BA.
The adsorption profiles of MN and ECA with the (nitrile) nitro-
gen atom adsorbed on the metal (Figs. 5b and 6b) resemble each
other and are in line with the ranking predicted in the preliminary
study with acetonitrile as probe molecule (Table 5). The energies of
the deprotonated MN or ECA donor molecules give a different
order, which takes into account all intrinsic properties of catalyst
and donor molecules (Figs. 5d and 6c). The deprotonation of MN
(Fig. 5d) leads to reaction energies which are slightly negative
(exothermic) with respect to the donor in gas phase. For Ni2dodbc,
the transition barrier for the deprotonation of MN was determined
to be 6 kJ/mol with respect to the deprotonated product state. As
the energy difference between transition state and deprotonated
product is so small, we can assume that the energy levels of the
deprotonated products are a good approximation of the deproto-
nating potential of Ni2dodbc. Provided the deprotonation step of
the donor molecule is rate determining for the overall reaction
and is thermodynamically driven, the sequence of the energy levels
of the deprotonated product may be regarded as an indicator for
the prediction of the ranking of the performance of the various M2-
dobdc catalysts in the Knoevenagel reaction. For the Knoevenagel
condensation of BA and MN on M2dobdc, the agreement is remark-
ably good, with the overall order of Ni > Cu > Mg ꢀ Zn > Co being
reflected in experiment (Table 1) and in theory (Fig. 5d).
Another plausible adsorption mode of the MN donor corre-
sponds to a double coordination with two terminating nitrile
groups on two adjacent metal sites, which geometrically fits per-
fectly for MN (Fig. 5c). As expected, the adsorption energies are
lower due to the energetically favorable interaction between the
nitrile groups and the metal centers. For instance, in the Ni2dobdc
case, the double coordination results in an adsorption energy that
is 45 kJ/mol more negative than the single coordination. This con-
firms the hypothesis that an excess of MN could block the coordin-
atively unsaturated sites, which could play a role in activating the
acceptor in Michael additions (see Fig. 2 and related discussion).
On the other hand, this double adsorption mode is subjected to a
larger entropic penalty as more degrees of freedom are immobi-
lized during this step, compensating for a large part the electronic
adsorption energies. We performed a frequency calculation for the
Ni case and found free energies of adsorption of À32 kJ/mol in case
of one adsorption site and À38 kJ/mol in case of adsorption of the
two nitrile groups at a temperature of 373 K. For the reaction to
proceed via deprotonation as in Fig. 5d, one of the coordinations
should be lifted, in order to allow the proton hopping to the pheno-
late oxygen of the framework.
In case of ECA interaction with an active site within M2dobdc,
ECA can be adsorbed and deprotonated in two different ways
(Fig. 6). First, we consider the adsorption of the nitrile group on
the metal site, followed by deprotonation. As ECA is less acidic than
MN, it is not surprising that a different order of activity is obtained
with positive reaction energies (endothermic) for most of the met-
als, except for Ni (Fig. 6c). These results are in line with the gener-
ally lower conversion rates observed experimentally for reactions
with ECA as donor after 2 h of reaction (Table 1). Second, the
adsorption of the (carbonyl) oxygen atom on a metal site looks also
an attractive pathway, as it results in a more stable adsorption
structure (Fig. 6d). However, after deprotonation, an end product
is formed that inhibits further reaction with benzaldehyde. The
proton for benzaldehyde adsorption is shielded and the C@C dou-
ble bond makes the reactive carbon less reactive (see Fig. 6e). The
formation of such a stable, but unreactive intermediate, would be
in line with the low experimentally observed overall rate for the
Knoevenagel reaction between ECA and BA (Table 1).
It is clear that besides the initial deprotonation and formation of
an adsorbed anionic intermediate, other steps in the catalytic cycle,
such as CAC coupling and dehydration (Scheme 2), could have
their impact on the relative overall rates observed for the various
M2dobdc catalysts. However, at least for the Knoevenagel conden-
sations, preliminary calculations have shown that the further con-
densation of the adsorbed anionic intermediate with BA, as shown
in Scheme 2, occurs in an almost barrierless fashion. Hence, for this

P. Valvekens et al. / Journal of Catalysis 317 (2014) 1–10
9
particular reaction, the subsequent elementary steps will not alter
the ranking of activity as extracted from the deprotonation process
of the donor adsorbed on the metal site of the framework (Fig. 5d).
For the Michael addition, such firm statements cannot yet be
made; deeper understanding may be acquired by modeling explic-
itly the reaction, likely including the adsorption of methyl vinyl
ketone on a Lewis acid site. Activation of both donor and acceptor
will also be affected by the equilibrium concentrations of the reac-
tants in the pores. Such overall computational study falls outside
the scope of the present theoretical calculation.
The only deviating case in all simulations is the Cu material, for
which the theory systematically predicts high energy intermedi-
ates, and hence, a very low reactivity. We have no plausible expla-
nation for this observation. The ground state valence electronic
configuration of Cu2+ is [Ar] 3d9 with one hole in the fully occupied
3d orbital (single electron in the axial 3d2z orbital). In contrast to
Mg2+ and Zn2+, both exhibiting fully occupied subshells, serious
distortions are noticed from the octahedral geometry in the opti-
mized Cu2dobdc unit cell. In the most stable state, we have found
that the unit cell of Cu2dobdc (containing 36 Cu-atoms) has a total
spin of about 1.9 and shows by far the smallest cell volume
(Table S.2) with very compressed CuACu distances (3 Å) in the
c-direction of the crystal. Probably, the LOT applied to the other
M2dobdc materials with success does not suit the Cu case. In this
context, the work of Nachtigall [72] on the role of the functional
in the prediction of adsorption energies of propane and propylene
on the MOF Cu3BTC2 is of particular interest.
project funding G.0453.09 and G.0486.12. P.V. and M.V. are grateful
for a fellowship from FWO Vlaanderen. Funding was also received
from the European Research Council under FP7 with ERC grant
agreement number 240483, the Ghent University and the FWO
Vlaanderen. Computational resources and services were provided
by Ghent University (Stevin Supercomputer Infrastructure).
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jcat.2014.06.006.
References
[1] J.R. Li, J. Sculley, H.C. Zhou, Chem. Rev. 112 (2012) 869–932.
[2] A. Corma, H. García, F.X. Llabrés i Xamena, Chem. Rev. 110 (2010) 4606–4655.
[3] P. Valvekens, F. Vermoortele, D. De Vos, Catal. Sci. Technol. 3 (2013) 1435–
1445.
[4] J. Gascon, U. Aktay, M.D. Hernandez-Alonso, G.P.M. van Klink, F. Kapteijn, J.
Catal. 261 (2009) 75–87.
[5] P. Kasinathan, Y.K. Seo, K.E. Shim, Y.K. Hwang, U.H. Lee, D.W. Hwang, D.Y.
Hong, S.B. Halligudi, J.S. Chang, Bull. Korean Chem. Soc. 32 (2011) 2073–2075.
[6] M. Savonnet, S. Aguado, U. Ravon, D. Bazer-Bachi, V. Lecocq, N. Bats, C. Pinel, D.
Farrusseng, Green Chem. 11 (2009) 1729–1732.
[7] Y. Tan, Z. Fu, J. Zhang, Inorg. Chem. Commun. 14 (2011) 1966–1970.
[8] F.X. Llabrés i Xamena, F.G. Cirujano, A. Corma, Microporous Mesoporous Mater.
157 (2012) 112–117.
[9] U.P.N. Tran, K.K.A. Le, N.T.S. Phan, ACS Catal. 1 (2011) 120–127.
[10] C. Chizallet, S. Lazare, D. Bazer-Bachi, F. Bonnier, V. Lecocq, E. Soyer, A.A.
Quoineaud, N. Bats, J. Am. Chem. Soc. 132 (2010) 12365–12377.
[11] L.T.L. Nguyen, K.K.A. Le, H.X. Truong, N.T.S. Phan, Catal. Sci. Technol. 2 (2012)
521–528.
[12] T. Maity, D. Saha, S. Das, S. Koner, Eur. J. Inorg. Chem. 2012 (2012) 4914–4920.
[13] D. Saha, T. Maity, R. Sen, S. Koner, Polyhedron 43 (2012) 63–70.
[14] D. Saha, R. Sen, T. Maity, S. Koner, Dalton Trans. 41 (2012) 7399–7408.
[15] R. Sen, D. Saha, S. Koner, Chem. Eur. J. 18 (2012) 5979–5986.
[16] P.D.C. Dietzel, Y. Morita, R. Blom, H. Fjellvåg, Angew. Chem., Int. Ed. 44 (2005)
6354–6358.
[17] P.D.C. Dietzel, B. Panella, M. Hirscher, R. Blom, H. Fjellvag, Chem. Commun.
(2006) 959–961.
[18] P.D.C. Dietzel, R. Blom, H. Fjellvåg, Eur. J. Inorg. Chem. 2008 (2008) 3624–3632.
[19] P.D.C. Dietzel, R.E. Johnsen, R. Blom, H. Fjellvåg, Chem. Eur. J. 14 (2008) 2389–
2397.
[20] W. Zhou, H. Wu, T. Yildirim, J. Am. Chem. Soc. 130 (2008) 15268–15269.
[21] N.L. Rosi, J. Kim, M. Eddaoudi, B. Chen, M. O’Keeffe, O.M. Yaghi, J. Am. Chem.
Soc. 127 (2005) 1504–1518.
[22] L. Mitchell, B. Gonzalez-Santiago, J.P.S. Mowat, M.E. Gunn, P. Williamson, N.
Acerbi, M.L. Clarke, P.A. Wright, Catal. Sci. Technol. 3 (2013) 606–617.
[23] E.D. Bloch, W.L. Queen, R. Krishna, J.M. Zadrozny, C.M. Brown, J.R. Long, Science
335 (2012) 1606–1610.
[24] S.J. Geier, J.A. Mason, E.D. Bloch, W.L. Queen, M.R. Hudson, C.M. Brown, J.R.
Long, Chem. Sci. 4 (2013) 2054–2061.
[25] D. Britt, H. Furukawa, B. Wang, T.G. Glover, O.M. Yaghi, PNAS 106 (2009)
20637–20640.
[26] P.D.C. Dietzel, V. Besikiotis, R. Blom, J. Mater. Chem. 19 (2009) 7362–7370.
[27] J. Liu, J. Tian, P.K. Thallapally, B.P. McGrail, J. Phys. Chem. C 116 (2012) 9575–
9581.
[28] E.J. García, J.P.S. Mowat, P.A. Wright, J. Pérez-Pellitero, C. Jallut, G.D. Pirngruber,
J. Phys. Chem. C 116 (2012) 26636–26648.
[29] S.M. Chavan, G.C. Shearer, E. Bloch, S. Bordiga, ChemPhysChem 13 (2012) 445–
448.
[30] P.K. Allan, P.S. Wheatley, D. Aldous, M.I. Mohideen, C. Tang, J.A. Hriljac, I.L.
Megson, K.W. Chapman, G. De Weireld, S. Vaesen, R.E. Morris, Dalton Trans. 41
(2012) 4060–4066.
[31] T. Remy, S.A. Peter, S. Van der Perre, P. Valvekens, D.E. De Vos, G.V. Baron, J.F.M.
Denayer, J. Phys. Chem. C 117 (2013) 9301–9310.
[32] S. Xiang, W. Zhou, Z. Zhang, M.A. Green, Y. Liu, B. Chen, Angew. Chem., Int. Ed.
49 (2010) 4615–4618.
[33] J.S. Lee, S.B. Halligudi, N.H. Jang, D.W. Hwang, J.S. Chang, Y.K. Hwang, Bull.
Korean Chem. Soc. 31 (2010) 1489–1495.
[34] Y. Fu, D. Sun, M. Qin, R. Huang, Z. Li, RSC Adv. 2 (2012) 3309–3314.
[35] S.R. Caskey, A.G. Wong-Foy, A.J. Matzger, J. Am. Chem. Soc. 130 (2008) 10870–
10871.
4. Conclusions
While the development of metal–organic frameworks as base
catalysts has mainly focused on the use of N-functionalized linkers,
the use of M2dobdc as a catalyst for Knoevenagel condensation or
Michael addition reactions shows that it is possible to use MOFs
with an intrinsic basic lattice activity. The catalytic activity of the
M2dobdc materials in these reactions can be ascribed to the pres-
ence of basic phenolate oxygen atoms in the structure, which,
together with the adjacent coordinatively unsaturated metal ions,
make up the active site. The nature of the metal ion has been
shown to strongly influence the overall activity, and these effects
were also investigated computationally. If, at least for the Knoeve-
nagel condensation, the deprotonation of the adsorbed donor mol-
ecules is the slowest step, and assuming that this first step is
thermodynamically driven, our computations succeed in predict-
ing a correct ranking of the activity for the metals under study. This
is remarkable, since such calculations do not take into account pos-
sible fluctuations of the numbers of well accessible active sites in
the M2dobdc materials, e.g. due to fault planes or channel obstruc-
tions in the crystals. Moreover, due to different adsorption prefer-
ences, the intraporous concentrations of the reactants may also
vary for the different materials. Finally, the role of the solvent
inside the pores was not considered.
There is a clear preference for Ni2dobdc as the generally most
active catalyst. The material with Ni at the nodal points combines
a clear basic activity, reflected in an energetically favorable depro-
tonation of ACH2A donor molecules, with sufficiently strong reac-
tant binding open metal sites, accounting for the generally
satisfactory catalytic performance displayed by Ni2dobdc.
[36] N. Trukhan, U. Muller, A. Panchenko, I. Malkowsky, A. Fischer, Novel metal–
organic frameworks as electrode material for lithium ion accumulators, in:
World Intellectual Property Organization (Ed.), BASF SE, 2011.
[37] J.L.C. Rowsell, O.M. Yaghi, J. Am. Chem. Soc. 128 (2006) 1304–1315.
[38] G. Kresse, J. Furthmüller, Phys. Rev. B 54 (1996) 11169–11186.
[39] G. Kresse, J. Furthmüller, Comput. Mater. Sci. 6 (1996) 15–50.
[40] G. Kresse, J. Hafner, Phys. Rev. B 47 (1993) 558–561.
[41] G. Kresse, J. Hafner, Phys. Rev. B 49 (1994) 14251–14269.
[42] J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865–3868.
Acknowledgments
The authors are grateful to the Belgian Federal Government for
support in the IAP project 07/05 Functional Supramolecular
Systems, D.V. and P.V. are grateful to K.U. Leuven for the
Methusalem CASAS Grant, and to FWO Vlaanderen for research

10
P. Valvekens et al. / Journal of Catalysis 317 (2014) 1–10
[43] J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 78 (1997). 1396-1396.
[44] S. Grimme, J. Antony, S. Ehrlich, H. Krieg, J. Chem. Phys. 132 (2010).
[45] P.E. Blöchl, Phys. Rev. B 50 (1994) 17953–17979.
[55] M. Shamzhy, M. Opanasenko, O. Shvets, J. Cejka, Front. Chem. 1 (2013).
[56] A. Corma, R.M. Martín-Aranda, J. Catal. 130 (1991) 130–137.
[57] C. Xu, J.K. Bartley, D.I. Enache, D.W. Knight, G.J. Hutchings, Synthesis 2005
(2005) 3468–3476.
[58] K. Mori, T. Hara, T. Mizugaki, K. Ebitani, K. Kaneda, J. Am. Chem. Soc. 125
(2003) 11460–11461.
[59] A. Coelho, Comb. Chem. High Throughput Screening 11 (2008) 843.
[60] P.J. Black, M.G. Edwards, J.M.J. Williams, Tetrahedron 61 (2005) 1363–1374.
[61] L. Valenzano, J.G. Vitillo, S. Chavan, B. Civalleri, F. Bonino, S. Bordiga, C.
Lamberti, Catal. Today 182 (2012) 67–79.
[46] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman,
G. Scalmani, V. Barone, B. Mennucci, G.A. Petersson, H. Nakatsuji, M. Caricato,
X. Li, H.P. Hratchian, A.F. Izmaylov, J. Bloino, G. Zheng, J.L. Sonnenberg, M.
Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y.
Honda, O. Kitao, H. Nakai, T. Vreven, J.J.A. Montgomery, J.E. Peralta, F. Ogliaro,
M. Bearpark, J.J. Heyd, E. Brothers, K.N. Kudin, V.N. Staroverov, R. Kobayashi, J.
Normand, K. Raghavachari, A. Rendell, J.C. Burant, S.S. Iyengar, J. Tomasi, M.
Cossi, N. Rega, J.M. Millam, M. Klene, J.E. Knox, J.B. Cross, V. Bakken, C. Adamo,
J. Jaramillo, R. Gomperts, R.E. Stratmann, O. Yazyev, A.J. Austin, R. Cammi, C.
Pomelli, J.W. Ochterski, R.L. Martin, K. Morokuma, V.G. Zakrzewski, G.A. Voth,
P. Salvador, J.J. Dannenberg, S. Dapprich, A.D. Daniels, Ö. Farkas, J.B. Foresman,
J.V. Ortiz, J. Cioslowski, D.J. Fox, Inc., Wallingford, CT, 2009.
[62] J.C. Lavalley, Catal. Today 27 (1996) 377–401.
ˇ
ˇ
[63] J. Kucera, P. Nachtigall, J. Kotrla, G. Košová, J. Cejka, J. Phys. Chem. B 108 (2004)
16012–16022.
[64] R.A. Jones, G.P. Bean, The Chemistry of Pyrroles, Academic Press, New York,
1977.
[47] A.D. Becke, J. Chem. Phys. 98 (1993) 5648–5652.
[65] P.O. Scokart, P.G. Rouxhet, J. Chem. Soc., Faraday Trans. 1 (76) (1980) 1476–
1489.
[66] S. Kobayashi, Y. Yamashita, Acc. Chem. Res. 44 (2010) 58–71.
[67] M. Hatano, T. Horibe, K. Ishihara, Angew. Chem., Int. Ed. (2013) 4549–4553.
[68] L. Valenzano, B. Civalleri, K. Sillar, J. Sauer, J. Phys. Chem. C 115 (2011) 21777–
21784.
[48] C. Lee, W. Yang, R.G. Parr, Phys. Rev. B 37 (1988) 785–789.
[49] T. Verstraelen, V. Van Speybroeck, M. Waroquier, J. Chem. Inf. Model. 48 (2008)
1530–1541.
[50] A. Ghysels, T. Verstraelen, K. Hemelsoet, M. Waroquier, V. Van Speybroeck, J.
Chem. Inf. Model. 50 (2010) 1736–1750.
[51] P. Serra-Crespo, E.V. Ramos-Fernandez, J. Gascon, F. Kapteijn, Chem. Mater. 23
(2011) 2565–2572.
[69] S. Chavan, F. Bonino, J.G. Vitillo, E. Groppo, C. Lamberti, P.D.C. Dietzel, A.
Zecchina, S. Bordiga, Phys. Chem. Chem. Phys. 11 (2009) 9811–9822.
[70] S. Chavan, J.G. Vitillo, E. Groppo, F. Bonino, C. Lamberti, P.D.C. Dietzel, S.
Bordiga, J. Phys. Chem. C 113 (2009) 3292–3299.
[71] L. Valenzano, B. Civalleri, S. Chavan, G.T. Palomino, C.O. Areán, S. Bordiga, J.
Phys. Chem. C 114 (2010) 11185–11191.
[52] M. Almasi, V. Zelenak, M. Opanasenko, J. Cejka, Dalton Trans. 43 (2014) 3730–
3738.
ˇ
[53] M. Opanasenko, A. Dhakshinamoorthy, J. Cejka, H. Garcia, ChemCatChem 5
(2013) 1553–1561.
[54] M. Opanasenko, A. Dhakshinamoorthy, M. Shamzhy, P. Nachtigall, M. Horacek,
H. Garcia, J. Cejka, Catal. Sci. Technol. 3 (2013) 500–507.
[72] M. Rubes, A.D. Wiersum, P.L. Llewellyn, L. Grajciar, O. Bludsky, P. Nachtigall, J.
Phys. Chem. C 117 (2013) 11159–11167.