Full Paper
space remains for catalysis in the yellow cavities, which do not
contain phosphine groups and therefore permit free diffusion
of relatively large molecules. 31P MAS NMR spectroscopy shows
that all P atoms in LSK-3 are accessible to react with reactants
(see below).
produced the equivalent amount of product after 120 h. Wash-
ing of the MOF with CHCl3 allowed reuse of the material in
a second cycle, where it mediated 89% conversion of reactants
into product (Figure 2A) compared to stoichiometric reaction
of PPh3.
Encouraged by these findings we probed the activity of LSK-
3 in umpolung addition of ethyl 2,3-butanedienoate to malo-
nonitrile yielding ethyl 5,5-dicyanopent-2-enoate (4; Fig-
ure 2B). This reaction, when catalyzed by PPh3 (20 mol%) in
the homogeneous phase, affords the product in 73% yield.[38]
LSK-3 (20 mol% P) did not produce even a trace of product
under identical conditions. In an effort to further probe the im-
pediment to this reaction we investigated the Knoevenagel
condensation of benzaldehyde with malononitrile to give 2-
benzylidene malononitrile (5; Figure 2C),[39] a reaction that is
also catalyzed by amino MOFs.[40] After 110 h reaction time at
808C in CHCl3, the conversion of benzaldehyde to product was
98%. The catalyst was recycled four times by consecutive
washings with dichloromethane leading to stabilized produc-
tivity at about 70% conversion. PXRD revealed that the struc-
ture remains intact after four recycling steps. Knoevenagel con-
densation with our P-MOF catalyst works three times faster
than the homogeneous reaction catalyzed by PPh3, which con-
verted only 37% of benzaldehyde after 110 h at 808C.
Catalytic Application of LSK-3
LSK-3 was tested as catalyst in four different types of stoichio-
metric and catalytic reactions with substrates of different type
and size. A summary of all catalytic results is shown in
Figure 2. In all reactions described below, we used samples
with relatively big single crystals of size 0.5–1.0 mm (as mea-
Finally, we treated the Knoevenagel product 5 with ethyl
2,3-butanedienoate and performed a [3+2] cycloaddition that
we hoped would yield ethyl 4,4-dicyano-5-phenylcyclopent-1-
enecarboxylate (6) as product (Figure 2D). Although the reac-
tion is successfully mediated by PPh3 catalyst (20 mol%),[41]
LSK-3 (20 mol% P) did not produce any product after 60 h of
reaction at 808C in CHCl3.
Molecular Modeling and Intermediate Characterization
Figure 2. Reactions catalyzed by LSK-3. A) Coumarin synthesis; run 1: >99%
conversion, run 2: 89% conversion. B) Umpolung addition. C) Knoevenagel
condensation; run 1: 98% conversion, run 2: 89% conversion, run 3: 73%
conversion, run 4: 72% conversion. D) [3+2] cycloaddition.
A number of MOF catalytic systems have shown size selectivity
induced by their pore size.[42–44] However, comparison of the
relative size of reactants and products (see the Supporting In-
formation, Table S1) to the available pore opening and void
space (Figure 1D,E) in LSK-3 excludes the possibility that tradi-
tional size-selectivity plays a role. In fact, the molecules partici-
pating in the umpolung reaction without any apparent conver-
sion are actually smaller than those reacting in the successfully
catalyzed coumarin synthesis. We therefore performed geome-
try optimization, to check the relative stability of the inter-
mediates oriented in a different way within the MOF cage, and
solid-state 31P NMR spectroscopy to confirm the calculation’s
results. Both analyses revealed that the orientation of reaction
intermediates enable or hamper reaction to occur (see below).
Geometric optimizations were performed with Materials Studio
from Accelrys with the Forcite molecular modeling (MM) opti-
mization, which takes account only for steric effects and not
sured by optical microscopy) to minimize the ratio between
the phosphine on the surface and that inside the pores. In this
way, we would indirectly prove that, if catalysis were to occur,
it would do so within the pores and not just on the surface of
the crystal. After all catalytic reactions, the material was recov-
ered and analyzed either by powder X-ray diffraction (PXRD) or
by polarized-light optical microscopy, which showed the single
crystal integrity in all catalytic attempts below (see the Sup-
porting Information, Figure S4). Notably, the non-functionalized
IRMOF-9 failed to catalyze any of the reactions in which LSK-3
was active. We explored the synthesis of coumarin methyl 6-
acetyl-2-oxo-2H-chromene-4-carboxylate (3) from 4-hydroxya-
cetophenone and dimethyl acetylenedicarboxylate (Figure 2A),
which is a reaction requiring a stoichiometric amount of tri-
phenylphosphine (PPh3) in the homogeneous phase.[37] LSK-3
(100 mol% P) effected quantitative conversion to the desired
coumarin 3 in CHCl3 after 100 h at 808C. By comparison, a reac-
tion carried out with PPh3 at an identical phosphorus loading
31
electronic ones, followed by DMol3 DFT gradient-corrected
(GGA) correlation functional of Perdew–Burke–Ernzerhof
(PBE)[46] geometry optimization with the precise numerical
basis set DNP 3.5 (double numerical plus polarization). Similar
DFT calculations have already been successfully used in MOF
geometry optimization.[47] The presence of one or two phos-
Chem. Eur. J. 2014, 20, 15467 – 15472
15469
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim