Synthesis modulation as a tool to increase the catalytic activity of metal-organic frameworks: The unique case of UiO-66(Zr)

Article abstract of DOI:10.1021/ja405078u

The catalytic activity of the zirconium terephthalate UiO-66(Zr) can be drastically increased by using a modulation approach. The combined use of trifluoroacetic acid and HCl during the synthesis results in a highly crystalline material, with partial substitution of terephthalates by trifluoroacetate. Thermal activation of the material leads not only to dehydroxylation of the hexanuclear Zr cluster but also to post-synthetic removal of the trifluoroacetate groups, resulting in a more open framework with a large number of open sites. Consequently, the material is a highly active catalyst for several Lewis acid catalyzed reactions.

Full text of DOI:10.1021/ja405078u

Communication
pubs.acs.org/JACS
Synthesis Modulation as a Tool To Increase the Catalytic Activity of
Metal−Organic Frameworks: The Unique Case of UiO-66(Zr)
Frederik Vermoortele,‡ Bart Bueken,‡ Gaelle Le Bars,§ Ben Van de Voorde,‡ Matthias Vandichel,†
̈
Kristof Houthoofd,‡ Alexandre Vimont,§ Marco Daturi,§ Michel Waroquier,† Veronique Van Speybroeck,†
Christine Kirschhock,‡ and Dirk E. De Vos*,‡
‡Centre for Surface Chemistry and Catalysis, Katholieke Universiteit Leuven, Arenbergpark 23, B-3001 Leuven, Belgium
§Laboratoire Catalyse et Spectrochimie, ENSICAEN, Universite
France
́ ́
de Caen, CNRS, 6 Boulevard du Marechal Juin, F-14050 Caen,
†Center for Molecular Modeling, Universiteit Gent, Technologiepark 903, B-9052 Zwijnaarde, Belgium
S
* Supporting Information
of Kitagawa, who used modulators as a tool for controlling size
and shape of [{Cu2(NDC)2(dabco)}]n and HKUST-1
6
ABSTRACT: The catalytic activity of the zirconium
terephthalate UiO-66(Zr) can be drastically increased by
using a modulation approach. The combined use of
trifluoroacetic acid and HCl during the synthesis results in
a highly crystalline material, with partial substitution of
terephthalates by trifluoroacetate. Thermal activation of
the material leads not only to dehydroxylation of the
hexanuclear Zr cluster but also to post-synthetic removal
of the trifluoroacetate groups, resulting in a more open
framework with a large number of open sites. Con-
sequently, the material is a highly active catalyst for several
Lewis acid catalyzed reactions.
crystals;7 however, earlier, similar examples can be found as
well.8 Besides organic carboxylic acids, also organic bases or
even inorganic modulators can exert analogous effects.8d,9 For
the zirconium dicarboxylate MOFs, modulators like acetic acid
were introduced by Schaatte et al.10 in the synthesis of UiO-66
and UiO-67. In these cases, the role of the modulator is
primarily to terminate the structure. However, a study of the
copper carboxylate MOF PCN-125 shows that a modulator
may also be incorporated inside the crystals. Thus, partial
replacement of the terphenyl-3,3″,5,5″-tetracarboxylate linker
by functionalized isophthalic acid resulted in the creation of
functionalized mesopores.11
Recently, we demonstrated that the unexpected catalytic
activity12 of the seemingly coordinatively saturated UiO-6613 is
due to missing linkers in the structure.14 Additionally, the
activity of the material can be significantly increased by using
electron-withdrawing groups on the terephthalate linkers.15
Here we demonstrate how the correct use of modulators, like
trifluoroacetic acid, can create even much more active materials.
UiO-66 was synthesized from a mixture containing
equimolar amounts of ZrCl4, terephthalic acid (BDC), and
H2O in N,N′-dimethylformamide (DMF), with addition of
different equivalents (X) of trifluoroacetic acid (TFA) (here-
after noted as UiO-66-X) with respect to the other reactants
(e.g., BDC). As modulators are expected to facilitate
crystallization for a variety of MOFs,7,8,10 it is not surprising
that the diffractograms show well-resolved signals, indicating
that the crystallinity of UiO-66-X is maintained, if not even
improved, when more TFA is added (Figure S1). Unlike
benzoic, acetic, or formic acid, however,10 TFA does not affect
the crystal size (Figure S3): use of TFA as a modulator leads to
the same morphology of small (∼300−600 nm) intergrown
particles as in the non-modulated synthesis.
esearchers have high expectations for metal−organic
Rframeworks (MOFs) in catalysis because of the high
concentration of metal ions dispersed in a micro- or
mesoporous matrix.1 However, the activity of current MOFs
rarely approaches that of other microporous solids, e.g.,
zeolites.2 One option to overcome this low activity is the
activation of dormant, fully coordinated framework metal ions,
which are unavailable for catalysis. While in some cases the
reactants seem to break the metal−linker bonds,3 the most
catalytically active MOFs typically remain those with structural
coordinatively unsaturated Lewis acid sites.1 Recently, several
tools were proposed to increase the catalytic activity of MOFs.
For instance, post-synthetic treatment of MIL-100(Fe) with
inorganic acids increases the number of Lewis and Bronsted
̈
acid sites, resulting in higher activity and selectivity in Diels−
Alder reactions.4 For MOF-5, Ravon et al. found that materials
prepared through fast precipitation were the most active ones in
alkylations, which was ascribed to a large number of OH groups
at structural defects.5 In this Communication, we show how the
use of specific synthesis modulators allows for a high level of
control over the number and nature of the defect sites in the
well-studied zirconium terephthalate MOF UiO-66 and results
in large increases of the catalytic activity.
The catalytic activity of UiO-66 synthesized with different
concentrations of modulator (0−20 equiv of TFA with respect
to BDC) was tested using the “ene”-type cyclization of
citronellal to isopulegol. All catalysts were activated under
In MOF preparation, a modulator has typically the same or a
similar chemical functionality as the linker molecules, but it is
monofunctional rather than bi- or polycipital. The term
“coordination modulation” was coined in 2009 by the group
Received: May 21, 2013
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja405078u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
vacuum at 320 °C for 12 h to ensure full dehydroxylation of the
inorganic cluster. From a screening of monocarboxylic acids
with different pKa’s, CF3COOH was found to be the modulator
giving by far the most active materials (see also Supporting
Information, Table S1). Conversion plots are given in Figure 1.
Figure 2. Conversion of citronellal with UiO-66-X synthesized with
and without HCl vs time (toluene, 110 °C, citronellal:Zr4+ = 15).
imbalance on the opposing cluster may be balanced by
(8+n)+
replacement of an OH− group of the [Zr6(OH)nO8−n
]
cluster (n = 4 for the ideal structure) by an O2− group, resulting
in n < 4. Because of the volatile and labile nature of TFA (bp =
72 °C), one may also remove the trifluoroacetate with a
thermal treatment, creating even more open sites.
Figure 1. Conversion of citronellal over UiO-66-X vs time (toluene,
110 °C, citronellal:Zr4+ = 15).
To validate such ideas, the model materials UiO-66-10HCl
and UiO-66HCl were properly characterized using thermogravi-
metric analysis (TGA), 19F NMR, FTIR, CD3CN chemisorp-
tion, and N2 physisorption. In the TGA of UiO-66HCl (Figure
S6), the weight loss between 400 and 500 °C is consistent with
a BDC:Zr6 ratio of 5, in line with the known linker deficiency
even in well-crystallized materials.14 The charge imbalance
created by the linker deficiencies could be compensated by Cl−
anions, as indicated by EDX measurements, or by changes in
the cluster itself (vide supra).17 The same BDC:Zr6 ratio of 5 is
also observed for the UiO-66 material synthesized without HCl.
This indicates that HCl alone does not result in the formation
of additional active sites, explaining why UiO-66 and UiO-66HCl
present the same catalytic activities (Figure 2).
By contrast, for the modulated material UiO-66-10HCl, TGA
shows a stepwise weight loss, and 19F solid-state NMR spectra
of materials pretreated at different temperatures help in
elucidating these steps. The as-synthesized material contains,
besides physisorbed TFA (−75 ppm), also trifluoroacetate
anions directly bound to the Zr cluster, with an intense, upfield-
shifted signal at −78 ppm (Figure 3). When UiO-66-10HCl is
The addition of small quantities of TFA to the synthesis
already significantly increases the activity; a maximum activity is
obtained using materials synthesized with 20 equiv of TFA.
While regular UiO-66 reaches a conversion of 34% after 10 h,
UiO-66-20 has converted almost 75% of citronellal after the
same time. The selectivity toward isopulegol, which directly
reflects Lewis acid strength,15 fluctuates for all (non)modulated
materials between 75 and 77%. This proves that the increase in
activity is due to the creation of a larger number of open sites,
rather than to a modified acid strength. The use of even larger
quantities of TFA during synthesis did not result in more active
materials.
In addition to the use of CF3COOH as a modulator, the
effect of adding HCl as a crystallizing agent was investigated.
The literature contains a wide variety of recipes to synthesize
UiO-66, with differences in concentrations, solvent, and with
addition of inorganic bases or acids. In the synthesis described
by Yang et al.,16 1 equiv of HCl is added to the synthesis
mixture to facilitate the upscaling. During crystal formation,
HCl has a double function: it slows down the hydrolysis of
ZrCl4 and counteracts the deprotonation of the dissolved
carboxylic acids. Consequently, in a mixture containing both
TFA and BDC, HCl favors the competitive incorporation of
trifluoroacetate. To assess the effect of added HCl on the
catalytic activity, UiO-66 and UiO-66-10 were synthesized with
the addition of 1 equiv of HCl (indicated as UiO-66HCl and
UiO-66-10HCl). Catalytic tests show that this addition of HCl to
the synthesis has a strong effect on the activity of the TFA
modulated material (Figure 2). The conversion of citronellal
after 1 h reaches almost 40% with UiO-66-10HCl, whereas UiO-
66-10 reaches a conversion of 21%. Strikingly, the use of HCl
has almost no effect on the activity of the non-modulated
materials.
To rationalize these observations, one needs to understand
the role of TFA during framework formation. The facile
deprotonation of the strongly acidic TFA suggests that
trifluoroacetate may be incorporated not only at the crystal
boundaries but also even on the hexanuclear cluster inside the
structure, instead of a terephthalate linker. The resulting charge
Figure 3. Mass-normalized 19F NMR spectra of (a) UiO-66-10HCl as
synthesized, (b) UiO-66-10HCl treated at 200 °C, and (c) UiO-66-
10HCl treated at 320 °C. Asterisks indicate spinning sidebands.
B
dx.doi.org/10.1021/ja405078u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
heated to 200 °C, the trifluoroacetate remains bound on the
cluster; the slight signal shift to −80 ppm and the sharper shape
of the signal are probably due to the removal of guest molecules
solvating the CF3 groups. When the material is activated at 320
°C, the fluorine signal is completely lost, revealing that the
trifluoroacetates are removed. Nevertheless, the structural order
in the material is fully retained (Figure S2).
In the TGA profile of UiO-66-10HCl, the corresponding loss
of TFA is discerned as a clear step between 270 and 320 °C.
Quantification suggests that, in UiO-66-10HCl treated at 200 °C,
the hexanuclear cluster is typically surrounded by one
CF3COO− and four terephthalate anions, with [Zr6O7(OH)]-
(BDC)4(CF3COO) as a plausible, correctly charge-balanced
formula. As no Cl− is observed in EDX (see SI), replacement of
OH− ions by O2− ions on the cluster is probably responsible for
the charge balance. Such a substantial incorporation of
CF3COO− proves that the modulator is present not only at
the periphery but also in the bulk of the crystal. The process of
trifluoroacetate incorporation in UiO-66-10HCl and its loss
upon heating was also followed by FTIR (Figure S7). The
concentration of OH groups (3670 cm−1) is significantly
statistically or, as suggested by Valenzano et al.,14 systematically
on a local but not global scale.
To empirically assess the number and strength of the Lewis
acid (LA) sites, CD3CN chemisorption was used, with in situ
activation at different temperatures prior to the chemisorption.
The spectra (Figure S8) show two well-defined signals, at 2260
cm−1 for physisorbed CD3CN and at 2298 cm−1 indicative for
CD3CN adsorbed on LA sites. The total number of LA sites at
different activation temperatures is given in Figure 4, taking
reduced for UiO-66-10HCl in comparison with UiO-66HCl
,
confirming that already before the thermal treatment, part of
the OH− ions are replaced by O2− groups to ensure the charge
balance. Following diagnostic bands during heating (e.g., at
1638 and ∼1200 cm−1 for CF3COO−) proves the gradual
disappearance of the −OH and trifluoroacetate groups. While
the dehydroxylation starts at a much lower temperature than
the CF3COO− removal, the onset of the modulator
disappearance coincides with the removal of the last portion
of hydroxyl groups (Figure S9). This is an indication that the
CF3COO− anions recombine with protons from the
hexanuclear cluster to allow departure of neutral CF3COOH,
leaving a material with [Zr6O8](BDC)4 as a formula.
To quantitatively support these findings, reaction free
energies for dehydroxylation and removal of CF3COOH were
obtained using state-of-the-art periodic DFT calculations with
the PBE functional in the VASP package, including D3-
dispersion corrections (see also SI). The reaction free energy
for dehydroxylation amounts to about −74 kJ/mol at 320 °C
and 10−3 mbar, indicating that the dehydroxylation is easily
achieved given the low free energy value. This value remains
largely the same for a Zr cluster on which TFA has been
incorporated. Removal of CF3COOH requires a free energy of
+24.7 kJ/mol, in line with the higher temperatures (320 °C)
needed experimentally for this process. For a partially
dehydroxylated cluster, the removal of CF3COOH is more
easily achieved, given the lower free energy of +2.9 kJ/mol.
These results indeed confirm that dehydroxylation occurs
readily, and that removal of CF3COOH becomes competitive
with water removal at higher activation temperatures.
With only 8 carboxylates surrounding the cluster, rather than
12 in the ideal structure, the question arises whether the loss of
these carboxylate groups occurs in an ordered way.20 High-
resolution XRD suggests the carboxylates are missing in a
randomized fashion over the length scale of multiple unit cells;
the cubic space group Fm3m still adequately describes the
diffraction patterns of materials in the desolvated state (a =
20.76 Å) or after activation at 320 °C (a = 20.69 Å). An
ordered linker removal, which still may occur for individual Zr
clusters, therefore does not result in a lowering of the overall
averaged symmetry which can be observed by XRD.14 Instead,
it can be concluded that linker removal occurs either fully
Figure 4. Number of Lewis acid sites as a function of activation
temperature.
into account a molar absorption coefficient of 3.1 cm/μmol for
coordinated species. This immediately shows that the use of
TFA strongly increases the number of LA sites in the material.
After an activation at 300 °C, UiO-66HCl has 720 μmol/g LA
sites, whereas nearly 1100 μmol/g LA sites are present in UiO-
66-10HCl. This corresponds to slightly more than two accessible
active sites per hexanuclear cluster.
Consecutive incorporation and loss of TFA not only results
in a significantly larger number of open metal sites but also
should influence the pore texture. For the non-modulated
material (Figure S10), the average pore size matches very well
the reported values; however, pore size drastically increases
when more TFA is incorporated. In the material with a
maximum amount of TFA incorporated (prepared with 20
equiv of TFA with respect to BDC), the pore structure has
completely changed from pores of originally 0.78 nm diameter
to a material with pores of 1−1.1 nm diameter.
The more open pore structure and the larger number of
open sites should make the modulated UiO-66 an excellent
catalyst for reactions in which several reactants need to be
activated simultaneously, or in which bulky reactants are to be
used. As an example, we studied the Meerwein reduction of 4-
tert-butylcyclohexanone (TCH) with isopropanol (IPA). This
reaction requires simultaneous activation of alcohol and ketone
to facilitate hydride transfer (Figure 5).
While the non-modulated materials display virtually no
activity (respectively 5 and 7% conversion after 24 h for UiO-
66 and UiO-66-NO2), the materials modulated with 10 equiv of
TFA display appreciable yields of the tert-butylcyclohexanol.
However, UiO-66-NO2 prepared using HCl and TFA is the
best catalyst, reaching 93% conversion. Clearly, the increased
space around the cluster facilitates the simultaneous activation
of the reactants, while the electronic effect of the nitro group is
also clear in this reaction.
C
dx.doi.org/10.1021/ja405078u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
(8) (a) Uemura, T.; Kitagawa, S. Chem. Lett . 2005, 34, 132.
(b) Uemura, T.; Hoshino, Y.; Kitagawa, S.; Yoshida, K.; Isoda, S.
Chem. Mater. 2006, 18, 992. (c) Cho, W.; Lee, H. J.; Oh, M. J. Am.
Chem. Soc. 2008, 130, 16943. (d) Wu, S.-T.; Wu, Y.-R.; Kang, Q.-Q.;
Zhang, H.; Long, L.-S.; Zheng, Z.; Huang, R.-B.; Zheng, L.-S. Angew.
Chem., Int. Ed. 2007, 46, 8475. (e) Hermes, S.; Witte, T.; Hikov, T.;
Zacher, D.; Bahnmuller, S.; Langstein, G.; Huber, K.; Fischer, R. A. J.
̈
Am. Chem. Soc. 2007, 129, 5324.
(9) Chin, J. M.; Chen, E. Y.; Menon, A. G.; Tan, H. Y.; Hor, A. T. S.;
Schreyer, M. K.; Xu, J. CrystEngComm 2013, 15, 654.
(10) Schaate, A.; Roy, P.; Godt, A.; Lippke, J.; Waltz, F.; Wiebcke,
M.; Behrens, P. Chem.Eur. J. 2011, 17, 6643.
(11) Park, J.; Wang, Z. U.; Sun, L.-B.; Chen, Y.-P.; Zhou, H.-C. J. Am.
Chem. Soc. 2012, 134, 20110.
(12) Vermoortele, F.; Ameloot, R.; Vimont, A.; Serre, C.; De Vos, D.
E. Chem. Commun. 2011, 47, 1521.
Figure 5. Conversion of TCH over UiO-66-X and UiO-66-NO2-X vs
time (toluene, 100 °C, TCH:IPA:Zr4+ = 10:50:1).
(13) Cavka, J. H.; Jakobsen, S.; Olsbye, U.; Guillou, N.; Lamberti, C.;
Bordiga, S.; Lillerud, K. P. J. Am. Chem. Soc. 2008, 130, 13850.
(14) Valenzano, L.; Civarelli, B.; Chavan, S.; Bordiga, S.; Nilsen, M.
H.; Jakobsen, S.; Lillerud, K. P.; Lamberti, C. Chem. Mater. 2011, 23,
1700.
(15) Vermoortele, F.; Vandichel, M.; Van de Voorde, B.; Ameloot,
R.; Waroquier, M.; Van Speybroeck, V.; De Vos, D. E. Angew. Chem.,
Int. Ed. 2012, 51, 4887.
(16) Yang, Q.; Jobic, H.; Salles, F.; Kolokolov, D.; Guillerm, V.;
Serre, C.; Maurin, G. Chem.Eur. J. 2011, 17, 8882.
(17) In this case, the proposed formula is [Zr6(OH)2O6](BDC)5];
charge balance around the cluster is achieved by partical replacement
of OH− ions by O2− anions. Note that analogous hexanuclear Zr6
clusters have been described as Zr6O8,18 Zr6(OH)4O4,13,14 or even
Zr6(OH)8,19 depending on the number of linker molecules bound.
(18) Worris, W.; Volosskiy, B.; Demir, S.; Gandara, F.; McGrier, P.
́
L.; Furukawa, H.; Cascio, D.; Stoddart, J. F.; Yaghi, O. M. Inorg. Chem.
2012, 51, 6443.
(19) Park, J.; Wang, Z. U.; Sun, L.-B.; Chen, Y.-P.; Zhou, H.-C. J. Am.
Chem. Soc. 2012, 134, 20110.
(20) The removal of TFA was also followed by monitoring the IR
spectra of the gases released during the heating of the sample from 200
to 320 °C, which showed that, besides desorption of intact
CF3COOH, the compound may also be decomposed with splitting
of the C−C bond.
In summary we have demonstrated that the catalytic activity
of zirconium terephthalates for a variety of reactions can be
enhanced by adding trifluoroacetic acid in the synthesis.
Trifluoroacetate groups replace part of the terephthalate linkers
in the material, creating extra Lewis acid sites and a more open
material. This approach can be highly valuable in exploiting the
full potential of MOFs as catalysts.
ASSOCIATED CONTENT
* Supporting Information
Synthesis procedures, experimental methods, physical charac-
terization, and computational procedures. This material is
available free of charge via the Internet at http://pubs.acs.org.
■
S
AUTHOR INFORMATION
Corresponding Author
dirk.devos@biw.kuleuven.be
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
D.E.D.V. is grateful to KULeuven for support in the
Methusalem grant CASAS, to IAP 7/05 (Belgium), and to
IWT (MOFShape). This work was further supported by FWO-
Vlaanderen, BOF-GOA(UGent), and ERC grant (no. 240483).
Computational resources were provided by Ghent University
(Stevin Supercomputer Infrastructure).
REFERENCES
(1) (a) Corma, A.; García, H.; Llabres
2010, 110, 4606. (b) Yoon, M.; Srirambalaji, R.; Kim, K. Chem. Rev.
2012, 112, 1196.
■
́
i Xamena, F. X. Chem. Rev.
(2) Dhakshinamoorthy, A.; Alvaro, M.; Corma, A.; García, H. Dalton
Trans. 2011, 40, 6344.
(3) Leus, K.; Vandichel, M.; Liu, Y.-Y.; Muylaert, I.; Musschoot, J.;
Pyl, S.; Vrielinck, H.; Callens, F.; Marin, G. B.; Detavernier, C.; Wiper,
P. V.; Khimyak, Y. Z.; Waroquier, M.; Van Speybroeck, V.; Van Der
Voort, P. J. Catal. 2012, 285, 196.
(4) Vermoortele, F.; Ameloot, R.; Alaerts, L.; Matthesen, R.; Gascon,
J.; Kapteijn, F.; De Vos, D. E. J. Mater. Chem. 2012, 22, 10313.
(5) Ravon, U.; Savonnet, M.; Aguado, S.; Domine, M. E.; Janneau, E.;
Farrusseng, D. Microporous Mesoporous Mater. 2010, 129, 319.
(6) Tsuruoka, T.; Furukawa, S.; Takashima, Y.; Yoshida, K.; Isoda, S.;
Kitagawa, S. Angew. Chem., Int. Ed. 2009, 48, 4739.
(7) (a) Diring, S.; Furukawa, S.; Takashima, Y.; Tsuruoka, T.;
Kitagawa, S. Chem. Mater. 2010, 22, 4531. (b) Umemura, A.; Diring,
S.; Furukawa, S.; Uehara, H.; Tsuruoka, T.; Kitagawa, S. J. Am. Chem.
Soc. 2011, 133, 15506.
D
dx.doi.org/10.1021/ja405078u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX