Phosphonate-Mediated Immobilization of Rhodium/Bipyridine Hydrogenation Catalysts

Article abstract of DOI:10.1002/chem.201705283

RhL₂ complexes of phosphonate-derivatized 2,2′-bipyridine (bpy) ligands L were immobilized on titanium oxide particles generated in situ. Depending on the structure of the bipy ligand—number of tethers (1 or 2) to which the phosphonate end groups are attached and their location on the 2,2′-bipyridine backbone (4,4′-, 5,5′-, or 6,6′-positions)—the resulting supported catalysts showed comparable chemoselectivity but different kinetics for the hydrogenation of 6-methyl-5-hepten-2-one under hydrogen pressure. Characterization of the six supported catalysts suggested that the intrinsic geometry of each of the phosphonate-derivatized 2,2′-bipyridines leads to supported catalysts with different microstructures and different arrangements of the RhL₂ species at the surface of the solid, which thereby affect their reactivity.

Full text of DOI:10.1002/chem.201705283

DOI: 10.1002/chem.201705283
Full Paper
&
Supported Catalysts
Phosphonate-Mediated Immobilization of Rhodium/Bipyridine
Hydrogenation Catalysts
Florian Forato,[a] Anouar Belhboub,[a] Julien Monot,[b] Marc Petit,[c] Roland Benoit,[d]
Vincent Sarou-Kanian,[e] Franck Fayon,[e] Denis Jacquemin,[a, f] ClØmence Queffelec,*[a] and
Bruno Bujoli[a]
Abstract: RhL2 complexes of phosphonate-derivatized 2,2’-
bipyridine (bpy) ligands L were immobilized on titanium
oxide particles generated in situ. Depending on the struc-
ture of the bipy ligand— number of tethers (1 or 2) to which
the phosphonate end groups are attached and their location
on the 2,2’-bipyridine backbone (4,4’-, 5,5’-, or 6,6’-posi-
tions)— the resulting supported catalysts showed compara-
ble chemoselectivity but different kinetics for the hydroge-
nation of 6-methyl-5-hepten-2-one under hydrogen pres-
sure. Characterization of the six supported catalysts suggest-
ed that the intrinsic geometry of each of the phosphonate-
derivatized 2,2’-bipyridines leads to supported catalysts with
different microstructures and different arrangements of the
RhL2 species at the surface of the solid, which thereby affect
their reactivity.
Introduction
control of the arrangement of the organic component at the
surface. Grafting of organometallic catalysts in a controlled and
stable manner requires the presence of functional groups that
vary according to the chemical nature of the solid support to
be modified. Some of the most popular combinations are
thiol/gold,[10] SiH3, Si(OR)3, or SiCl3/silica or glass,[11–13] and
P(O)(OR)2 (R=H or alkyl)/metal oxides[14–25] or any type of ma-
terials containing surface-exposed metal centers.[26]
Grafting molecular catalysts onto surfaces to form heterogene-
ous systems offers several potential advantages, including
easier product separation and catalyst recovery. Common
routes for immobilizing molecular catalysts include their incor-
poration into organic polymers[1] and their adsorption onto
high surface area inorganic frameworks such as zeolites and
mesoporous micelle-templated materials, silica, or other metal
oxides.[2–9] Another strategy is derivatization of solid materials
with molecular catalysts, which consists of grafting appropriate
functional molecules onto the surface and provides improved
Immobilized catalysts usually mirror the behavior of their ho-
mogeneous counterparts, but the accessibility of the catalytic
site can be altered by the orientation of the catalyst with re-
spect to the surface or by aggregation, which results in lower
reaction rates. Interestingly, confinement effects related, for ex-
ample, to the embedment of organometallic catalysts in meso-
porous solids[27,28] or ordered surfaces (self-assembled mono-
layers,[26,29–33] dendrimers[34–36]) can lead to unusual catalytic be-
havior.
[a] F. Forato, Dr. A. Belhboub, Prof. D. Jacquemin, Dr. C. Queffelec, Dr. B. Bujoli
Chimie Et InterdisciplinaritØ: Synth›se Analyse ModØlisation (CEISAM)
University of Nantes, CNRS, UMR 6230
2, rue de la Houssini›re, BP 92208, 44322 Nantes Cedex 3 (France)
E-mail: clØmence.queffelec@univ-nantes.fr
The use of rhodium complexes in catalytic hydrogenation re-
actions is reported in some contributions,[37,38] and in particular
rhodium/2,2’-bipyridine complexes have been shown to
reduce C=N, C=O, C=C, and CꢀC bonds. Previously, we have in-
vestigated the immobilization of a rhodium complex based on
phosphonate-functionalized 2,2’-bipyridine A (Scheme 1) on ti-
tanium oxide particles generated in situ.[15,37] The resulting sup-
ported organometallic catalyst was shown to reduce ketones
efficiently.[15] However, in the case of the competitive hydroge-
nation of ketones versus carbon–carbon multiple bonds under
hydrogen pressure in alkaline methanol, the supported catalyst
was more active for the reduction of C=C and CꢀC bonds than
for the reduction of keto groups, in contrast to the analogous
homogeneous system, for which the opposite was observed.[37]
However, the precise reason for this difference was not eluci-
dated.
[b] Dr. J. Monot
CNRS, UniversitØ Paul Sabatier
Laboratoire HØtØrochimie Fondamentale et AppliquØe (LHFA, UMR 5069)
118 Route de Narbonne, 31062 Toulouse Cedex 09 (France)
[c] Dr. M. Petit
Sorbonne UniversitØs, UPMC UNIV Paris 06
Institut Parisien de Chimie MolØculaire, UMR CNRS 8232, Case 229
4 Place Jussieu, 75252 Paris Cedex 05 (France)
[d] Dr. R. Benoit
ICARE-CNRS
1C Avenue de la Recherche Scientifique, 45100 OrlØans (France)
[e] Dr. V. Sarou-Kanian, Dr. F. Fayon
CNRS, CEMHTI UPR3079, Univ. OrlØans, 45071 OrlØans (France)
[f] Prof. D. Jacquemin
Institut Universitaire de France
1, rue Descartes, 75231 Paris Cedex 05 (France)
Supporting information for this article can be found under:
https://doi.org/10.1002/chem.201705283.
Chem. Eur. J. 2018, 24, 2457 –2465
2457
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Results and Discussion
Synthesis of 2,2’-bipyridine ligands functionalized with
phosphonate end groups
The location and number of phosphonate tethers on the 2,2’-
bipyridine backbone were expected to induce differences in
the orientation and accessibility of the corresponding rhodium
complexes when bound to the metal oxide surface. According-
ly, six ligands were prepared having one or two phosphonate
tethers located at different positions of the 2,2’-bipyridine
rings. Ligands mono-4-C12, mono-5-C12, mono-6-C12, 4,4’-
C12, 5,5’-C12, and 6,6’-C12 were prepared by condensation of
the appropriate 2,2’-bipyridine mono- or dicarbaldehyde 1–6
with diethyl oxyaminododecylphosphonate (7) in ethanol, fol-
lowed by hydrolysis of the phosphonate esters by the McKen-
na method (Scheme 2).[39] For more details of all syntheses, see
the Supporting Information.
Scheme 1. Immobilization of rhodium/bipyridine catalysts on titanium oxide
by means of phosphonate chemistry.[37]
The purpose of the present work was thus 1) to prepare a
series of phosphonate-functionalized 2,2’-bipyridines of differ-
ent geometry (Figure 1), in which the number (1 or 2) and lo-
cation of phosphonate tethers on the 2,2’-bipyridine backbone
(4,4’-, 5,5’-, or 6,6’-positions) were varied, and 2) to investigate
the potential influence of the geometry of these ligands on
the catalytic behavior of the resulting supported rhodium com-
plexes.
Scheme 2. General synthesis of the 2,2’-bipyridine-based mono- and bis-
phosphonic acids in this study.
Figure 1. Phosphonate-functionalized 2,2’-bipyridines prepared herein.
Preparation of the rhodium/2,2’-bipyridine catalysts
To this end, the 2,2’-bipyridine backbone was functionalized
with aldehyde group(s) to allow attachment of phosphonate
tethers bearing oxyamine end groups (NH2O(CH2)12P(O)(OR)2)
In the first step, [Rh(COD)Cl]2 (COD=1,5-cyclooctadiene) was
treated with AgBF4 in THF for 16 h[40,41] to form a more reactive
species, namely, [Rh(COD)(solvent)2]+BF4À, after removal of the
AgCl precipitate. The resulting solution was then mixed with
an aqueous solution of mono-4-C12, mono-5-C12, mono-6-
C12, 4,4’-C12, 5,5’-C12, or 6,6’-C12 (water/THF=1:1, v/v, Rh/
ligand=1:2), and the mixture was stirred for 2 d for complete
formation of the expected RhL2 complexes. Then, immobiliza-
tion of the resulting complexes was performed by a previously
optimized heterogenization protocol,[15] based on the reaction
of the phosphonic acid groups on the ligands with titanium
oxide particles generated in situ (Scheme 1). Typically, Ti(OiPr)4
was added in one portion to the above solution, with immedi-
ate formation of a poorly organized hydrated TiOx(OH)4À2x
compound and concomitant grafting of the catalytic complex
on its surface. Even though the hydrolysis/condensation of tita-
nium isopropoxide is very rapid under our experimental condi-
tions, entrapment of a small part of the rhodium/ bipyridine
complex in the oxide matrix cannot be ruled out. After 2 d, the
through the formation of
a stable O-alkyloxime bond
(ÀC=NOÀ). A flexible dodecyl tether was selected for this study
to minimize steric constraints around the corresponding immo-
bilized rhodium/bipyridine complexes. Successful immobiliza-
tion of the catalysts on titanium oxide particles was demon-
strated by different characterization techniques [31P magic-
angle spinning (MAS) NMR spectroscopy, X-ray photoelectron
spectroscopy (XPS), and time-of-flight secondary-ion mass
spectrometry (ToF-SIMS)]. The catalytic behavior of the result-
ing six supported catalysts in the hydrogenation of 6-methyl-5-
hepten-2-one under hydrogen pressure was found to be simi-
lar, with selectivity in favor of the reduction of C=C bonds.
However, significant differences in the reaction rate were ob-
served, depending on the number and location of the phos-
phonate tether(s) on the ligand, which influenced the micro-
structure of the supported catalysts.
Chem. Eur. J. 2018, 24, 2457 –2465
www.chemeurj.org
2458
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
resulting pale yellow solid was separated by centrifugation,
and the filtrate was analyzed by the Ames method to deter-
mine the residual amount of phosphorus, which allowed deter-
mination, by difference, of the amount of phosphorus in the
solid phase.[42] In all cases, the amount of residual phosphorus
Characterization of the supported catalysts
The chemical composition of the six supported catalysts was
investigated with various techniques. The P/Rh ratio was first
determined through measurement of the phosphorus and rho-
in the supernatant was close to zero and evidenced that 92– dium content by XPS and SEM/energy-dispersive X-ray (EDX)
99% of the rhodium/bipyridine complex (Table 1) was efficient-
ly bound to the titanium oxide.
analyses. The latter was found to be consistent with the immo-
bilization of RhL2 catalytic species (Table 2). In addition, the
XPS spectra showed an expected doublet corresponding
to Rh 3d5/2 and Rh 3d3/2, at binding energies of about 309
and 314 eV, respectively, close to that reported for the
[Cp*RhIII(bpy)Cl]+ complex[47] or rhodium oxide.[48]
Table 1. Amount of rhodium/ bipyridine complex immobilized in the
supported catalysts.
Ligand
Immobilization
yield [%]
Amount of bipy ligand
loaded on the supported catalyst
[mmolgÀ1
]
Table 2. Experimental P/Rh ratio measured for the supported catalysts by
SEM-EDX or XPS.
mono4-C12
mono5-C12
mono6-C12
4,4’-C12
5,5’-C12
6,6’-C12
99
98.5
96
94.5
92
93
0.52
0.51
0.50
0.49
0.48
0.48
Supported catalyst
Exptl (calcd) P/Rh ratio
mono4-C12
mono5-C12
mono6-C12
4,4’-C12
2.3 (2)[a]
2.0 (2)[b]
1.8 (2)[b]
3.9 (4)[b]
3.8 (4)[a]
3.6 (4)[a]
5,5’-C12
6,6’-C12
To facilitate the preparation, handling, and characterization
of the supported catalysts, their synthesis was not performed
under inert atmosphere; hence, rhodium was present in oxida-
tion state 3+, since [RhI(bpy)2]+ species are known to be
highly air sensitive.[43] By using 2,2’-bipyridine, a rhodium com-
plex was prepared in methanol according to the same proto-
col, which led to a pale yellow solution that on hydrogenation
(40 bar, 4 h) underwent a color change to reddish purple. This
was consistent with the formation of [RhI(bpy)2]+ species,
which are known to feature an intense UV/Vis metal-to-ligand
charge-transfer (MLCT) band around 550 nm with a weaker ab-
sorption at about 515 nm,[43–46] and rapid discoloration oc-
curred on exposure of this solution to air (Figure 2), due to fast
reoxidation. The similar color change that was observed for
the six supported catalysts on treatment with hydrogen
showed that formation of the [RhI(ligand)2]+ catalytic active
species also proceeds efficiently under these conditions.
[a] Measured by SEM-EDX. [b] Measured by XPS.
Complete characterization by elemental analysis (C, H, N)
and ICP-MS (Na, P, Rh, Ti) was also performed on the 6,6’-C12
supported catalyst (see Experimental Section) in order to pro-
pose an approximate formula for this material:
TiO1.5(OH)·3H2O·0.3iPrOH·Na0.33·[Rh(PO3(CH2)12ON=CH-bpy-CH=
NO(CH2)12PO3)2]0.04. In the case of complete immobilization of
the rhodium/bipyridine complex in the titanium oxide matrix,
the calculated Rh/Ti molar ratio is 0.044. In the present case
the experimental value of this ratio corresponds to about 90%
of the complex effectively immobilized on the supported cata-
lyst, in full agreement with the measurement reported in
Table 1.
ToF-SIMS[49] was then used to confirm grafting of the rhodi-
um/2,2’-bipyridine complexes onto the titanium dioxide sur-
face. For example, in the case of the mono-5-C12-based sup-
ported catalyst, the negative-ion mass spectrum (with a bis-
muth source) showed relevant signals assigned to [PO2]À,
[PO3]À, [TiO3H]À, [RhNC2H4]À, [RhNC3H2]À, [PO2Ti3H]À, [PO3Ti3]À,
[PO3Ti3O]À, and [N2C9OH11]À cluster ions (see Figure 3, top).
In addition, the positive-ion mass spectrum gave evidence
of
[Rh]+,
[RhN2]+,
[RhNC5H4]+,
[RhNC6H6]+,
and
[RhN2C6OH5(CH2)n]+ (n=0, 1, 2, 3 and 4) cluster ions (see
Figure 3, bottom). All of these data are consistent with cova-
lent attachment of the rhodium/bipyridine species to the tita-
nium oxide matrix through the formation of P-O-Ti bonds.
For the series of immobilized catalysts, the binding modes
of the phosphonic acid anchors on the titanium oxide surface
were investigated by solid-state 31P MAS NMR spectroscopy.
The obtained quantitative 31P MAS NMR spectra exhibit broad
resonances due to distributions of 31P chemical shifts, as ex-
pected for phosphonate groups linked to a disordered titani-
Figure 2. UV/Vis absorption spectra of a rhodium/2,2’-bipyridine reference
solution before (black curve), immediately after (reddish purple curve), and
10 min after (yellow curve) hydrogenation.
Chem. Eur. J. 2018, 24, 2457 –2465
www.chemeurj.org
2459
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Table 3. 31P isotropic chemical shifts (dISO Æ0.3 ppm), full width at half
maximum (FWHMÆ0.2 ppm), and relative intensities (IÆ3%) of the dis-
tinct contributions used for the simulation of the 31P quantitative MAS
NMR spectra of the supported catalysts. For the mono4-C12, mono5-
C12, and 5,5’-C12 catalysts the chemical-shift anisotropy (dCSA Æ3 ppm)
and asymmetry parameter (hCSA Æ0.1) determined from 2D PASS spectra
(2 kHz spinning frequency) are included.
Sample
Line
dISO
FWHM
[ppm]
I
dCSA
hCSA
[ppm]
[%]
[ppm]
mono-4-C12
1
2
3
1
2
3
1
2
3
4
1
2
3
1
2
3
4
1
2
3
4
25.3
29.0
34.0
25.3
29.8
34.3
25.6
29.7
34.0
27.5
25.3
30.0
33.5
25.2
29.0
34.0
26.1
25.5
29.0
34.0
26.1
9.1
6.1
7.0
8.4
5.1
6.5
7.8
6.5
6.5
2.0
9.0
6.0
6.0
8.8
7.0
6.0
1.5
8.3
7.0
6.0
1.5
65
27.5
7.5
56
31
13
60
35
2.5
2.5
67
27
6
À34
30
29
0.9
0.9
0.8
0.9
0.9
0.7
mono-5-C12
mono-6-C12
À33
Figure 3. ToF-SIMS spectra of mono-5-C12 (Bi3 cluster source). Top: nega-
tive-ion mass spectrum. Bottom: positive-ion mass spectrum.
30
28
um oxide surface (Figure 4). In most cases, asymmetric line
shapes were clearly observed and suggested the presence of
several overlapping contributions. Accordingly, all 31P MAS
spectra were reconstructed by considering three broad lines,
4,4’-C12
5,5’-C12
65
25
9
À34
32
30
0.8
0.9
0.8
1
6,6’-C12
75
22
2.5
0.5
CSA parameters associated with a given isotropic shift can be
determined (Figure 5). The 2D 31P slow-spinning-rate PASS
spectra recorded for several catalysts (mono4-C12, mono5-
C12, and 5,5’-C12) revealed differences in the CSA parameters
associated with the three broad peaks at about 25, 29.5, and
34 ppm (Table 3), and thus confirmed the presence of distinct
31P local chemical environments.
To assign the distinct contributions in 31P MAS spectra, DFT
computations of 31P chemical shifts were performed. For
simplification, propylphosphonic acid and an anatase surface
were used. As demonstrated previously in high-field 17O MAS
NMR studies,[51] different stable binding modes of phosphonic
acids can be expected on anatase. Our studies have been fo-
Figure 4. Left: 31P quantitative MAS NMR spectra of the six supported cata-
lysts. The asterisks indicate spinning sidebands. Right: expansion of the
center-band region (black lines) of the a) mono4-C12, b) mono5-C12,
c) mono6-C12, d) 4,4’-C12, e) 5,5’-C12, and f) 6,6’-C12 catalysts and their
best fits (red lines). The corresponding individual contributions are shown
below each spectrum.
with the most predominant one centered at dISO ꢁ25 ppm and
the two others at about 29 and 34 ppm (Figure 4, Table 3). For
the mono-6-C12, 4,4’-C12, and 6,6’-C12 materials, addition of
a narrow peak (at 26 or 27.5 ppm) of very weak intensity
(<2.5% of the total intensity) allowed better reconstruction of
the experimental spectra. This narrow minor peak is likely asso-
ciated with ungrafted species and is neglected in the following
discussion.
The presence of three distinct broad contributions in the
MAS spectra was also supported by 31P chemical shift anisotro-
py (CSA) measurements. To this end, a 2D PASS experiment[50]
was used to separate, according to their order, all the spinning
sidebands that overlap in slow-spinning-rate (2 kHz) MAS spec-
tra. This allows, after a tilt transformation, 2D correlation spec-
tra between the isotropic MAS spectrum and arrays of spin-
ning-sideband intensities to be obtained, from which average
Figure 5. a) 31P 2D (tilted) PASS spectra of the mono5-C12 sample recorded
at a spinning frequency of 2 kHz. b) Cross sections displaying spinning-side-
band intensities associated with isotropic chemical shifts of 23, 30, and
36 ppm and their corresponding best fits.
Chem. Eur. J. 2018, 24, 2457 –2465
www.chemeurj.org
2460
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
cused on the (101) surface, which is usually predominant in
anatase, and it has been shown that mono- and bidentate an-
choring is possible on this surface, whereas the tridentate
binding mode is not favored.[52] For a monodentate binding
mode, three different conformations can occur, depending on
the number of protons (two, one, or none) on the phosphonic
acid group, while for bidentate anchoring two configurations
are possible (Figure 6).
Table 4. Binding energies [eV] of the different arrangements considered
in this study.
Parameter
Tilted
Parallel
Perpendicular
Emb onodentate (a)
Ebbidentate (d)
À0.99
À0.71
À0.75
À0.60
À0.84
À0.48
Table 5. Binding energies of the different arrangements considered in
this study. See Figure 7 for the configurations.
(a)
(b)
(c)
(d)
(e)
Eb [eV]
À0.99
1.04
À2.25
À0.71
À2.25
dergo hydrogen bonding with water molecules and/or hydrox-
yl groups present on the titanium oxide surface, but the explic-
it description of this hydrated surface is very difficult.
The 31P chemical shift tensor was then calculated for all the
energetically possible adsorption complexes (see Table S1 in
the Supporting Information). 31P isotropic chemical shifts and
CSA parameters referenced to H3PO4 are given in Table 6.
Figure 6. Representation of grafted molecules of propylphosphonic acid on
a anatase TiO2 (101) surface in monodentate and bidentate binding modes
exhibiting different degrees of deprotonation.
Table 6. Calculated 31P isotropic chemical shifts and CSA parameters for
the stable monodentate and bidentate binding modes for a tilted orien-
tation of the molecule with respect to the surface.
The orientation of the phosphonate molecule relative to the
surface was also considered. The cases of tilted (ca. 458), paral-
lel, and perpendicular orientations of the molecule (Figure 7)
with respect to the surface were compared for monodentate
and bidentate conformations (Figure 6a and d, respectively) by
calculation of the binding energies Eb =Emol+slabÀEmolÀEslab. The
obtained values, which are listed in Table 4, show that the
tilted orientation is the most stable for both mono- and biden-
tate anchoring. Accordingly, the binding energies of the other
possible conformations were calculated with a tilted orienta-
tion of the molecule with respect to the surface to investigate
whether they were energetically possible. The results (Table 5)
show that configuration b is energetically unstable, while con-
figurations c and e were found to be the most stable. In reality,
the phosphonic acid groups in configurations c and e likely un-
Configuration monodentate a monodentate c bidentate d bidentate e
dISO [ppm]
dCSA [ppm]
hCSA
18.2
147
0.1
26.1
À13
0
13.2
158
0.2
29.4
12
0
These values are in good agreement with previous calcula-
tions.[52] On this basis, the 31P chemical shifts obtained from fits
of the experimental spectra of the supported rhodium/ bipyri-
dine complexes suggest that the monodentate binding
mode c is predominant (line at ca. 25 ppm). Moreover, given
that tridentate binding on the anatase (001) surface was re-
ported to yield an isotropic chemical shift close to 25 ppm,[52]
small contributions from this binding mode cannot be defini-
tively ruled out, even though the abundance of this surface is
usually low. Finally, the line at ca. 29.5 ppm might be assigned
to the concomitant presence of the bidentate binding mode e
to a lesser extent, while conformations a and d are excluded.
This assignment is also consistent with 31P CSA parameters,
though the calculated values are smaller than the experimental
ones for conformations c and e. However, the real situation is
more complex, since under our experimental conditions hy-
drolysis of titanium isopropoxide leads to the formation of a
poorly organized hydrated TiOx(OH)4À2x compound. Hence,
there is no “pure” (101) or (001) surface, and one should be
cautious in establishing a direct correlation between theoreti-
cal and experimental 31P chemical shifts to determine the bind-
ing modes of the phosphonate anchors. Nevertheless, the
NMR data confirm that immobilization of the rhodium/bipyri-
dine complexes results in multiple binding modes of the phos-
Figure 7. From left to right: tilted, parallel, and perpendicular configurations
of grafted propylphosphonic acid on the TiO2 anatase (101) surface for mon-
odentate (top) and bidentate (middle) binding modes. For clarity, a front
view of the bidentate anchoring is also shown (bottom).
Chem. Eur. J. 2018, 24, 2457 –2465
www.chemeurj.org
2461
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
phonate groups on the metal oxide surface. The observed
broad line widths also suggest a distribution of immobilized
species with different orientations with respect to the metal
oxide surface, which differ according to the number and loca-
tion of the phosphonate tether(s) on the ligand.
Catalytic activity of the supported catalysts compared to
the homogeneous reference system
6-Methyl-5-hepten-2-one was chosen as the substrate for the
assessment of the catalytic activity of the supported catalysts
and their selectivity for C=C versus C=O reduction. Preactiva-
tion of the catalysts (40 bar, 4 h) was performed before addi-
Finally, the microstructures of the six supported catalysts
were compared. SEM images showed that they consist of ag-
gregated particles of small size in the range of about 100– tion of the substrate. The same reaction time (i.e. 40 h) was
150 nm (see Figure S1 in the Supporting Information). Nitrogen
adsorption/desorption experiments were performed on the
whole series of supported catalysts, and the BET surface areas,
BJH desorption average pore diameters, and BJH desorption
cumulative volumes of pores between 2 and 57.5 nm diameter
are given in Table 7.
used to investigate whether the kinetics of the reaction was in-
fluenced by the structure of the immobilized ligand. The ho-
mogeneous reference cationic catalyst based on 2,2’-bipyridine
and [Rh(COD)]BF4 showed high selectivity for the reduction of
C=C bonds and complete conversion (Table 8, entry 1; see also
Supporting Information). The opposite selectivity was previous-
ly reported for the homogeneous neutral catalyst based on the
[{Rh(COD)Cl}2] precursor.[37]
Table 7. Results of the nitrogen adsorption/desorption studies on the six
supported catalysts.
The behavior of the six supported rhodium/bipyridine com-
plexes investigated under the same conditions showed in
many cases lower conversion rates than the homogeneous ref-
erence system. The highest conversion were obtained for
mono-6-C12 (99% conversion after 40 h; Table 8, entry 4)
among the monofunctionalized ligands, and for 4,4’-C12 (85%
conversion after 40 h; Table 8, entry 6) among the bis-function-
alized ligands. If required, full conversion can be obtained by
increasing the reaction time, as evidenced in Table 8 (entries 4
and 5). In terms of selectivity, reduction of the C=C bond large-
ly predominated for the series of supported catalysts, and the
best performance was observed for 4,4’-C12 with 100% selec-
tivity for hydrogenation of the olefin. One experiment (ligand:
mono-6-C12, reaction time: 40 h) was performed on a larger
scale to measure the yield of isolated product (95%), and the
composition of the reaction mixture was analyzed by GC and
1H NMR spectroscopy. The major constituent measured by
both methods was 6-methylheptan-2-one (85%), with small
amounts of 6-methyl-5-hepten-2-ol (5%), 6-methylheptan-2-ol
(5%), and remaining starting material (5%). These results are
thus consistent with the data of entry 4 in Table 8.
Catalyst
BET surface area BJH desorption BJH desorption
[m2 gÀ1
]
average pore
diameter [nm]
cumulative pore
volume (2–57.5 nm)
[cm3 gÀ1
]
mono-4-C12
mono-5-C12
mono-6-C12
4,4’-C12
5,5’-C12
6,6’-C12
122Æ2
198Æ2
106Æ2
103Æ2
58.5Æ0.5
150Æ1
6.8
5.7
11.8
11.6
10.9
8.9
0.12
0.17
0.11
0.08
0.12
0.30
Depending on the structure of the 2,2’-bipydine ligand, the
six heterogenized complexes showed different surface areas
(ranging from ca. 60 to 200 m2 gÀ1), average pore diameters
(ranging from ca. 6 to 12 nm), and porous volumes (ranging
from ca. 0.08 to 0.3 cm3 gÀ1). This demonstrates that the nature
of the ligand significantly influences the microstructure of the
resulting supported catalysts, which might in turn affect their
respective catalytic performances.
Interestingly, the fastest kinetics of the catalytic reaction
(4,4’-C12 and mono-6-C12) correspond to supported catalysts
showing the largest average pore diameter (ca. 11–12 nm),
Table 8. Results of the hydrogenation of 6-methyl-5-hepten-2-one with the homogeneous reference system and the supported catalysts.[a]
Entry
Catalyst
Reaction time [h]
Conversion rate [%]
Distribution of the reaction products [%]
1[b]
2
3
4
5
6
7
8
reference
40
40
40
40
24
40
40
40
99
15
40
99
70
85
73
47
90
90
70
85
90
100
93
87
10
–
15
–
–
–
7
–
10
15
15
10
–
–
–
mono4-C12
mono5-C12
mono6-C12
mono6-C12
4,4’-C12
5,5’-C12
6,6’-C12
13
[a] All reactions were carried out in alkaline methanol at 308C under 40 bar H2 with a catalyst/substrate molar ratio of 1%. GC analysis of the filtered reac-
tion medium allowed the relative amounts of 6-methyl-5-hepten-2-one (substrate), 6-methylheptan-2-one, 6-methyl-5-hepten-2-ol, and 6-methylheptan-2-
ol to be quantified, from which the substrate conversion and the distribution of reaction products were determined; no other product was detected.
[b] Experiment performed under homogeneous conditions with 2,2’-bipyridine and [Rh(COD)]BF4.
Chem. Eur. J. 2018, 24, 2457 –2465
www.chemeurj.org
2462
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
while the slowest kinetics (mono-4-C12 and mono-5-C12) cor-
respond to supported catalysts having the smallest average
pore diameter (ca. 6–7 nm). This suggests that the reaction
rate is strongly dependent on the microstructure of the sup-
ported catalyst, which itself varies according to the structure of
the immobilized ligand. Moreover, the arrangement of the rho-
dium/bipyridine complex on the surface of the oxide matrix is
likely to be different as a function of the immobilized ligand,
and this might also affect the catalytic performance. Finally, to
verify the stability of the supported catalysts, catalytic runs
were carried out under our classical experimental conditions
(40 bar H2, 40 h, in alkaline methanol), except that the support-
ed catalyst was replaced by a bipyridine phosphonate precur-
sor in its phosphonic acid form. These blank experiments were
carried out for two different bipyridine ligands. At the end of
the reaction, the bipyridine ligand was quantitatively recovered
in both cases, and the corresponding NMR spectra did not
show any trace of cleavage of the oxime group.
droxide particles generated in situ was found to be influenced
by the strategy used for their grafting on the inorganic sup-
port. When one or two dodecylphosphonate tethers branched
via a C=NÀO connector were introduced at the 4,4’-, 5,5’-, or
6,6’-positions of the 2,2’-bipyridine backbone, the hydrogena-
tion of 6-methyl-5-hepten-2-one under hydrogen pressure oc-
curred preferentially at the C=C bond, consistent with the che-
moselectivity found for the homogenous parent catalyst. How-
ever, the manner in which the phosphonate end groups are in-
troduced into the 2,2’-bipyridine backbone (number of tethers
and their location on the aromatic rings) was found to strongly
influence the microstructure of the resulting supported cata-
lysts (surface area, pore size distribution, etc.), which, in turn,
affected the reaction rates. The accessibility and the arrange-
ment of the phosphonate-functionalized catalytic complex at
the surface of the inorganic support is therefore dependent on
its structure, but it is difficult to predict a priori which structure
will be optimal in terms of catalytic performances. It would
thus be interesting to investigate whether different results are
obtained on immobilization of the rhodium complexes of our
six phosphonate-derivatized 2,2’-bipyridine ligands on a pre-
formed mesoporous titanium dioxide support.
Recycling experiments
The ability of the supported catalysts to be reused was investi-
gated for 4,4’-C12 (Table 9). After the first cycle, the catalyst
was separated by filtration from the reaction mixture, washed
twice with methanol, dried, and reused under similar condi-
tions. A slow decrease in yield was observed after the second
reuse, and led to 60% conversion after five cycles, due to
losses during the isolation of the catalyst by filtration leading
to a decrease of the catalyst/substrate ratio. Indeed, the cata-
lytic reactions were performed on a small scale (i.e., ca. 40 mg
of catalyst loaded in the reactor) and the amount of catalyst
recovered by filtration after the fourth cycle was about 25 mg,
while the amount of substrate was not adjusted to take ac-
count of the catalyst losses on the filter membrane during the
successive filtrations. The extent of rhodium leaching was de-
termined by inductively coupled plasma atomic emission spec-
troscopy (ICP-AES) analysis of the upper phase after each cata-
lytic cycle, which showed that the percentage of rhodium re-
leased in the solvent was relatively low, likely resulting from
rhodium decomplexation.
Experimental Section
General chemistry
All commercially available reagents and HPLC-grade solvents were
purchased from Aldrich, Acros, or Alfa Aesar. THF was distilled from
sodium/benzophenone. The protocols for the synthesis of the
mono-4-C12, mono-5-C12, mono-6-C12, 4,4’-C12, 5,5’-C12, and
6,6’-C12 ligands are detailed in the Supporting Information. Previ-
ously described methods were used for the preparation of
À
[40,41]
[Rh(bipy)2]+BF4
.
1
Solution-state H, 13C, and 31P spectra were recorded in deuterated
solvents with a Bruker Avance spectrometer operating at 300 or
400 MHz. Solid-state 31P MAS NMR experiments were performed
with a Bruker Avance III spectrometer operating at 9.4 T (1H and
31P Larmor frequencies of 400 and 161.9 MHz, respectively) by
using a 4 mm double-resonance MAS probe. Quantitative 31P MAS
spectra were recorded at a spinning frequency of 10 or 14 kHz
with a pulse length of 0.9 ms (228 flip angle) and a recycle delay of
1 s. 2D PASS experiments were performed at a spinning frequency
of 2 kHz. The 908 pulse duration was 3.7 ms (31P nutation frequency
of 67.5 kHz) and the recycle delay was set to 20 s. 31P chemical
shits were referenced relative to 85% H3PO4. Line-shape recon-
structions and spinning-sideband intensity analyses were per-
Conclusion
The reactivity of [RhL2]+ rhodium complexes of phosphonate-
derivatized 2,2’-bipyridines immobilized on titanium oxide/hy-
Table 9. Results from the hydrogenation of 6-methyl-5-hepten-2-one with the 4,4’-C12 supported catalyst.
Cycle
Conversion [%]
Rhodium loss in the supernatant [%][a]
[%]
[%]
[%]
1
2
3
4
5
85
85
80
75
60
100
100
95
90
90
–
–
–
5
–
–
5
5
5
0.4
1.8
1.8
1.5
1.5
5
[a] Amount of rhodium released in solution with respect to the total amount of rhodium initially present in the supported catalyst.
Chem. Eur. J. 2018, 24, 2457 –2465
www.chemeurj.org
2463
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
formed with the DMFit program.[53] ToF-SIMS experiments were
performed with a TOF.SIMS 5 instrument from ION-TOF GmbH by
using a 25 keV bismuth liquid metal ion gun (LMIG). The primary
ion beam of 25 keV Bi3+ was operated in a static mode with a pri-
mary pulsed beam current of 2.5 pA. Appropriate charge neutrali-
zation (20 eV electronic flood gun) was used to avoid charging of
the samples during data collection. The detection was performed
in the mass range from 1 to 800 mu. With a data acquisition time
of 100 s, the total fluence did not reach 1012 ionscmÀ2, ensuring
static conditions. The XPS measurements were performed with a
Thermo Scientific ESCALAB 250 by using monochromatic AlKa
(1486.6 eV) radiation with a spot size of about 600 mm. The high-
resolution spectra were collected with 20 eV pass energy at a take-
off angle of q=908, and the C 1s peak position of 284.6 eV was
used to calibrate the spectra. After Shirley background subtraction,
XP spectra were deconvoluted into Gaussian–Lorentzian compo-
nents.
Immobilization of catalysts on titanium oxide-based parti-
cles
AgBF4 (0.06 mmol) was added to a solution of [Rh(COD)Cl]2
(14.8 mg; 0.03 mmol) in THF (5 mL).[40] The mixture was stirred
overnight and then filtered through a Millipore apparatus to
remove AgCl. The filtrate was then added to the desired ligand
(0.12 mmol) suspended in 5 mL of water, and the mixture was
stirred for 48 h at room temperature. Then titanium tetraisoprop-
oxide (0.4 mL) and 1m sodium hydroxide (0.5 mL) were added and
the mixture was stirred for two additional days. The reaction
medium was then centrifuged, and the collected solid was washed
four times with distilled water in centrifugation tubes and then
twice with acetone. After filtration, the supported catalysts were al-
lowed to dry at room temperature, and an amount of 400–450 mg
of solid was typically obtained, depending on the nature of the
ligand used (i.e., mono- or bis-functionalized). From the amount of
rhodium/bipyridine complex immobilized on each supported cata-
lyst (see Table 1), the calculated rhodium loading was about
0.25 mmol Rh per gram of catalyst. A full analysis of the chemical
composition was performed for one of the supported catalysts by
elemental analysis (C, H, N) and ICP-MS (Na, P, Rh, Ti). The 6,6’-C12
compound was selected, and the experimental values allowed an
approximate formula for this material to be proposed:
TiO1.5(OH)·3H2O·0.3iPrOH·Na0.33·[Rh(PO3(CH2)12ON=CH-bpy-CH=
Qualitative elemental analysis by SEM-EDX was performed with a
JEOL JSM 5800 LV scanning electron microscope with SAMx SDD
EDS instrument. ICP-AES was performed with a Thermo Fisher Sci-
entific iCAP 6300 instrument. For the calibration curve, solutions at
different concentrations of rhodium were prepared from a com-
mercial ICP standard solution of Rh at 1000 mgmLÀ1 in a 1% aque-
ous solution of nitric acid.
NO(CH2)12PO3)2]0.04. The resulting experimental and calculated
values are the following: Ti (calcd 20.7%, exptl 20.6%), Rh (calcd
1.8%, exptl 1.7%), Na (calcd 3.3%, exptl 3.3%), P (calcd 2.2%, exptl
2.3%), N (calcd 1.9%, exptl 1.9%), C (calcd 19.6%, exptl 18.8%), H
(calcd 6.0%, exptl 5.5%).
Computational studies
First-principles calculations were performed in the framework of
DFT by using the VASP code.[54] The generalized gradient approxi-
mation (GGA) was adopted to treat the exchange-correlation term
with the Perdew, Burke, and Ernzerhof (PBE) functional.[55] Projected
augmented wave (PAW) potentials were used to describe core
electrons.[56] During the geometry optimization, valence electrons
were expanded with a plane-wave basis set by using a 400 eV
energy cutoff. Dispersion corrections were included by using DFT-
D2,[57] that is, an empirical dispersion term was directly added to
the Kohn–Sham energies. Atomic positions of all systems were op-
timized with a total energy threshold of 10À3 eV while taking into
account only the G-point for the Brillouin zone sampling. For the
TiO2 surface, a (101) anatase slab with two stacked O-Ti-O layers
was considered. This thickness was proven to be sufficient to
model anatase surfaces.[58,59] The slab was built with dimensions of
21”19 Š in the xy plane. The size of the slab was chosen to pro-
vide sufficient separation between periodic images when the opti-
mized isolated molecule was adsorbed, in order to avoid spurious
interactions. Indeed, the adsorbate–adsorbate distance was greater
than 14 Š in the x and y directions, while the minimum adsorbate–
surface distance was 24 Š in the z direction. The structural optimi-
zations of the slab for the bare surface and hybrid systems were
performed by moving the atoms of the top layer in direction (001)
only, while the bottom layer was kept fixed. The molecules were
chemisorbed onto the surface by bonding its oxygen atoms to the
unsaturated titanium ions of the surface. Depending on the
degree of coordination, the dissociated proton(s) from the mole-
cule was(were) transferred to the surface and bonded to an
oxygen atom with a distance of 8 Š from the adsorption site in
order to prevent any influence on the intersystem interaction. The
chemical-shift calculations were performed with a higher 500 eV
energy cutoff for the plane-wave basis set. A denser 2”2”1 Mon-
khorst–Pack k-points grid was also adopted for these calculations.
The shifts were calculated by considering only the valence contri-
bution; the core contribution was kept frozen.
General procedure for the hydrogenation of alkenes
The appropriate amount of catalyst (i.e., ca. 40 mg corresponding
to 9.5 mmol of rhodium, vide supra) was transferred to a 30 mL
glass-coated stainless steel autoclave. Methanol (3 mL), 1m aque-
ous sodium hydroxide solution (47.5 mL; NaOH/Rh=5) were then
added. The autoclave was purged six times with argon and then
three times with hydrogen, and the final H2 pressure was adjusted
to 40 bar. The mixture was stirred for 4 h in a sand bath at 308C.
After this time, the pressure of the autoclave was released, and 6-
methyl-5-hepten-2-one (0.95 mmol, rhodium/substrate ratio:
1.0 mol%) was added. Then, the autoclave was purged six times
with argon and then three times with hydrogen, and the final H2
pressure was adjusted to 40 bar. The mixture was stirred in a sand
bath at 308C for the desired reaction time and then centrifuged.
The liquid phase was then isolated and analyzed by gas chroma-
tography to determine the conversion, the nature of the reaction
products, and their yields. These measurements were performed
with a Hewlett-Packard HP 6890 chromatograph with a J&W Scien-
tific DB-1 column (l=30 m, i.d.=0.25 mm, film thickness=
0.25 mm), equipped with
a FID detector (H2 as carrier gas
(1.5 mLminÀ1); temperature program: 608C (1 min), 808C
(2.58CminÀ1). Retention times: 6-methylheptan-2-one (5.25 min), 6-
methylheptan-2-ol (5.72 min), 6-methyl-5-hepten-2-one (6.05 min),
6-methyl-5-hepten-2-ol (6.43 min).
Acknowledgements
P. Janvier is gratefully acknowledged for his help with the ICP-
AES measurements and analysis of the phosphorus content in
solution using the Ames method. F. X. Lef›vre is gratefully ac-
knowledged for his help with the SEM observation of the sup-
Chem. Eur. J. 2018, 24, 2457 –2465
www.chemeurj.org
2464
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[26] I. O. Benítez, B. Bujoli, L. J. Camus, C. M. Lee, F. Odobel, D. R. Talham, J.
Am. Chem. Soc. 2002, 124, 4363–4370.
[27] M. Bonnet, L. Schmid, A. Baiker, F. Diederich, Adv. Funct. Mater. 2002, 12,
39–42.
[28] F. Goettmann, C. Sanchez, J. Mater. Chem. 2007, 17, 24–30.
[29] K. Hara, S. Tayama, H. Kano, T. Masuda, S. Takakusagi, T. Kondo, K.
Uosaki, M. Sawamura, Chem. Commun. 2007, 4280–4282.
[30] A. Pasc-Banu, C. Sugisaki, T. Gharsa, J.-D. Marty, I. Gascon, M. Krämer, G.
Pozzi, B. Desbat, S. Quici, I. Rico-Lattes, Chem. Eur. J. 2005, 11, 6032–
6039.
[31] A. Pasc-Banu, C. Sugisaki, T. Gharsa, J.-D. Marty, I. Gascon, G. Pozzi, S.
Quici, I. Rico-Lattes, C. Mingotaud, Angew. Chem. Int. Ed. 2004, 43,
6174–6177; Angew. Chem. 2004, 116, 6300–6303.
[32] M. W. McKittrick, C. W. Jones, Chem. Mater. 2005, 17, 4758–4761.
[33] K. Tollner, R. Popovitz-Biro, M. Lahav, D. Milstein, Science 1997, 278,
2100–2102.
[34] D. Astruc, F. Chardac, Chem. Rev. 2001, 101, 2991–3024.
[35] A.-M. Caminade, P. Servin, R. Laurent, J.-P. Majoral, Chem. Soc. Rev. 2008,
37, 56–67.
ported catalysts. The authors are grateful to ICMN in Orleans
for access to its surface analysis instruments. The authors are
grateful to Institut des MatØriaux Jean Rouxel in Nantes (S.
Grolleau) for his help for the BET surface area and BJH pore
size distribution measurements. A.B. acknowledges the Europe-
an Research Council (ERC, Marches 278845 Grant) for support-
ing his postdoctoral position. D.J. acknowledges the European
Research Council (ERC) and the RØgion des Pays de la Loire for
financial support in the framework of a Starting Grant (Marches
278845) and the LumoMat project, respectively. This research
used computational resources of (i) the GENCI-CINES/IDRIS; (ii)
CCIPL (Centre de Calcul Intensif des Pays de Loire); (iii) a local
Troy cluster; and (iv) HPC resources from ArronaxPlus (Grant
ANR-11-EQPX-0004 funded by the French National Agency for
Research).
[36] P. Servin, R. Laurent, L. Gonsalvi, M. Tristany, M. Peruzzini, J.-P. Majoral,
A.-M. Caminade, Dalton Trans. 2009, 4432–4434.
Conflict of interest
[37] C. Maillet, P. Janvier, M. J. Bertrand, T. Praveen, B. Bujoli, Eur. J. Org.
Chem. 2002, 1685–1689.
[38] P. Frediani, L. Rosi, L. Cetarini, M. Frediani, Inorg. Chim. Acta 2006, 359,
2650–2657.
The authors declare no conflict of interest.
[39] C. E. McKenna, M. T. Higa, N. H. Cheung, M. C. McKenna, Tetrahedron
Lett. 1977, 18, 155–158.
[40] P. E. A. Ribeiro, C. L. Donnici, E. N. dos Santos, J. Organomet. Chem.
2006, 691, 2037–2043.
Keywords: hydrogenation · immobilization · N ligands
rhodium · supported catalysts
·
[41] D. Amarante, C. Cherian, C. Ernmel, H. Y. Chen, S. Dayal, M. Koshy, E. G.
Megehee, Inorg. Chim. Acta 2005, 358, 2231–2238.
[42] B. N. Ames, in Methods in Enzymology, Vol. VIII: Complex Carbohydrates.
(Eds.: E. Neufeld and V. Ginsburg), Academic Press, New York, 1966,
pp. 115–118.
[1] N. E. Leadbeater, M. Marco, Chem. Rev. 2002, 102, 3217–3274.
[2] A. Corma, H. García, A. Leyva, Tetrahedron 2004, 60, 8553–8560.
[3] R. Raja, J. M. Thomas, M. D. Jones, B. F. G. Johnson, D. E. W. Vaughan, J.
Am. Chem. Soc. 2003, 125, 14982–14983.
[4] M. J. Sabater, A. Corma, A. Domenech, V. FornØs, H. García, Chem.
Commun. 1997, 1285–1286.
[43] S. G. Yan, B. S. Brunschwig, C. Creutz, E. Fujita, N. Sutin, J. Am. Chem.
Soc. 1998, 120, 10553–10554.
[5] L. Zhang, T. Sun, J. Y. Ying, Chem. Commun. 1999, 1103–1104.
[6] Q.-H. Fan, Y.-M. Li, A. S. C. Chan, Chem. Rev. 2002, 102, 3385–3466.
[7] C. A. McNamara, M. J. Dixon, M. Bradley, Chem. Rev. 2002, 102, 3275–
3300.
[44] C. E. Castillo, M. Gennari, T. Stoll, J. Fortage, A. Deronzier, M.-N. Collomb,
M. Sandroni, F. Legalite, E. Blart, Y. Pellegrin, J. Phys. Chem. C 2015, 119,
5806–5818.
[45] M. T. Indelli, C. Chiorboli, F. Scandola, in Photochem. Photophysics Coord.
Compd. I (Eds.: V. Balzani, S. Campagna), Springer-Verlag Berlin, Berlin,
2007, pp. 215–255.
[46] E. Fujita, B. S. Brunschwig, C. Creutz, J. T. Muckerman, N. Sutin, D.
Szalda, R. van Eldik, Inorg. Chem. 2006, 45, 1595–1603.
[47] L. Zhang, N. Vilà, G.-W. Kohring, A. Walcarius, M. Etienne, ACS Catal.
2017, 7, 4386–4394.
[48] A. B. Kroner, M. A. Newton, M. Tromp, A. E. Russell, A. J. Dent, J. Evans,
ChemPhysChem 2013, 14, 3606–3617.
[8] R. Duchateau, Chem. Rev. 2002, 102, 3525–3542.
[9] M. Benaglia, A. Puglisi, F. Cozzi, Chem. Rev. 2003, 103, 3401–3430.
[10] T. Belser, M. Stçhr, A. Pfaltz, J. Am. Chem. Soc. 2005, 127, 8720–8731.
[11] D. E. De Vos, M. Dams, B. F. Sels, P. A. Jacobs, Chem. Rev. 2002, 102,
3615–3640.
[12] C. E. Song, S. Lee, Chem. Rev. 2002, 102, 3495–3524.
[13] A. P. Wight, M. E. Davis, Chem. Rev. 2002, 102, 3589–3614.
[14] A. Hu, H. L. Ngo, W. B. Lin, Angew. Chem. Int. Ed. 2003, 42, 6000–6003;
Angew. Chem. 2003, 115, 6182–6185.
[49] J. Grams, J. Goralski, P. Kwintal, Int. J. Mass Spectrom. 2010, 292, 1–6.
[50] O. N. Antzutkin, S. C. Shekar, M. H. Levitt, J. Magn. Reson. Ser. A 1995,
115, 7–19.
[15] C. Maillet, P. Janvier, M. Pipelier, T. Praveen, Y. Andres, B. Bujoli, Chem.
Mater. 2001, 13, 2879–2884.
[16] H. L. Ngo, A. G. Hu, W. B. Lin, J. Mol. Catal. A 2004, 215, 177–186.
[17] T. L. Schull, L. Henley, J. R. Deschamps, R. J. Butcher, D. P. Maher, C. A.
Klug, K. Swider-Lyons, W. J. Dressick, B. Bujoli, A. E. Greenwood, Organo-
metallics 2007, 26, 2272–2276.
[51] F. Brodard-Severac, G. Guerrero, J. Maquet, P. Florian, C. Gervais, P. H.
Mutin, Chem. Mater. 2008, 20, 5191–5196.
[52] D. Geldof, M. Tassi, R. Carleer, P. Adriaensens, A. Roevens, V. Meynen, F.
Blockhuys, Surf. Sci. 2017, 655, 31–38.
[18] D. Deniaud, G. A. Spyroulias, J. F. Bartoli, P. Battioni, D. Mansuy, C. Pinel,
F. Odobel, B. Bujoli, New J. Chem. 1998, 22, 901–905.
[19] M. Tassi, A. Roevens, G. Reekmans, M. Vanhamel, V. Meynen, J. D’Haen,
P. Adriaensens, R. Carleer, Adv. Powder Technol. 2017, 28, 236–243.
[20] C. QueffØlec, M. Petit, P. Janvier, D. A. Knight, B. Bujoli, Chem. Rev. 2012,
112, 3777–3807.
[21] A. Roevens, J. G. Van Dijck, M. Tassi, J. D’Haen, R. Carleer, P. Adriaensens,
F. Blockhuys, V. Meynen, Mater. Chem. Phys. 2016, 184, 324–334.
[22] L. Zeininger, L. Portilla, M. Halik, A. Hirsch, Chem. Eur. J. 2016, 22,
13506–13512.
[23] B. Andrews, S. Almandali, K. James, S. Ly, K. N. Crowder, Polyhedron
2016, 114, 360–369.
[53] D. Massiot, F. Fayon, M. Capron, I. King, S. Le Calve, B. Alonso, J. O.
Durand, B. Bujoli, Z. H. Gan, G. Hoatson, Magn. Reson. Chem. 2002, 40,
70–76.
[54] G. Kresse, J. Furthmuller, Comput. Mater. Sci. 1996, 6, 15–50.
[55] J. P. Perdew, K. Burke, Y. Wang, Phys. Rev. B 1996, 54, 16533–16539.
[56] P. Blçchl, Phys. Rev. B 1994, 50, 17953–17979.
[57] S. Grimme, J. Comput. Chem. 2006, 27, 1787–1799.
[58] F. De Angelis, S. Fantacci, E. Mosconi, M. K. Nazeeruddin, M. Graetzel, J.
Phys. Chem. C 2011, 115, 8825–8831.
[59] N. Martsinovich, D. R. Jones, A. Troisi, J. Phys. Chem. C 2010, 114,
22659–22670.
[24] C. Schmitt Pauly, A.-C. Genix, J. G. Alauzun, J. Jestin, M. Sztucki, P. H.
Mutin, J. Oberdisse, Polymer 2016, 97, 138–146.
[25] A. Mitrofanov, S. Brandes, F. Herbst, S. Rigolet, A. Bessmertnykh-Le-
meune, I. Beletskaya, J. Mater. Chem. A 2017, 5, 12216–12235.
Manuscript received: November 7, 2017
Accepted manuscript online: November 27, 2017
Version of record online: January 25, 2018
Chem. Eur. J. 2018, 24, 2457 –2465
www.chemeurj.org
2465
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim