Nb-doped variants of high surface aluminium fluoride: A very strong bi-acidic solid catalyst

Article abstract of DOI:10.1039/c9dt00831d

A niobium doped high surface aluminium fluoride (HS-AlF₃) catalyst was prepared, using an approach in which niobium doped aluminium hydroxide fluoride obtained via reaction of aqueous HF with the respective metal alkoxides in isopropanol is further fluorinated under flow of CHClF₂ at 200 °C. A comparable procedure was used to synthesize a Nb-free variant for comparison. Both catalysts exhibit very strong Lewis acidic surface sites which are capable to activate strong carbon-halogen bonds at room temperature, just as the classical high-surface AlF₃ (HS-AlF₃), obtained by reacting aluminium isopropoxide with anhydrous HF, does. The catalysts were characterized by elemental analysis, P-XRD, MAS NMR spectroscopy, N₂ adsorption, NH₃-TPD, and pyridine photoacoustic FT-IR spectroscopy. In contrast to previously reported niobium doped HS-AlF₃, which was prepared using anhydrous HF, the doped catalyst obtained via this aqueous HF-route shows excellent performance both in the isomerization of 1,2-dibromohexafluoropropane, a reaction that occurs only in the presence of the strongest Lewis acids, and in the cyclization of citronellal to isopulegol, a reaction which requires both, Lewis and Br?nsted acid sites.

Full text of DOI:10.1039/c9dt00831d

Dalton
Transactions
View Article Online
View Journal
PAPER
Nb-doped variants of high surface aluminium
fluoride: a very strong bi-acidic solid catalyst†‡
Cite this: DOI: 10.1039/c9dt00831d
Clara Patricia Marshall,a,b Gudrun Scholz,a Thomas Braun *a and
a
Erhard Kemnitz
*
A niobium doped high surface aluminium fluoride (HS-AlF3) catalyst was prepared, using an approach in
which niobium doped aluminium hydroxide fluoride obtained via reaction of aqueous HF with the
respective metal alkoxides in isopropanol is further fluorinated under flow of CHClF2 at 200 °C. A com-
parable procedure was used to synthesize a Nb-free variant for comparison. Both catalysts exhibit very
strong Lewis acidic surface sites which are capable to activate strong carbon–halogen bonds at room
temperature, just as the classical high-surface AlF3 (HS-AlF3), obtained by reacting aluminium isopropox-
ide with anhydrous HF, does. The catalysts were characterized by elemental analysis, P-XRD, MAS NMR
spectroscopy, N2 adsorption, NH3-TPD, and pyridine photoacoustic FT-IR spectroscopy. In contrast to
previously reported niobium doped HS-AlF3, which was prepared using anhydrous HF, the doped catalyst
obtained via this aqueous HF-route shows excellent performance both in the isomerization of 1,2-dibromo-
hexafluoropropane, a reaction that occurs only in the presence of the strongest Lewis acids, and in the
cyclization of citronellal to isopulegol, a reaction which requires both, Lewis and Brønsted acid sites.
Received 24th February 2019,
Accepted 16th April 2019
DOI: 10.1039/c9dt00831d
rsc.li/dalton
The fluorolytic sol–gel approach for metal fluorides allows
for flexibility in the synthesis procedure, since many para-
Introduction
Metal fluorides can be regarded as an interesting alternative to meters can be changed independently, such as concentration,
their metal oxide counterparts. Fluorine has a higher electro- temperature, metal precursor source, solvent, etc.4,5 The pres-
negativity than oxygen, and therefore, metal fluorides should ence of water plays a crucial role, since hydrolysis of the
usually exhibit a higher Lewis acidity than the respective metal–alkoxide bond of the aluminium alkoxide precursor
oxides. Moreover, metal fluorides are thermally and chemically may be an issue if water is added to the reaction mixture, thus
stable as well.1,2 More specifically, aluminium fluoride-based creating additional surface Brønsted sites. Consequently, in
catalysts have attracted attention since decades, in particular the past years, several studies have been published where the
after the development of the fluorolytic sol–gel synthesis in effect of water addition was studied for metal fluorides pre-
2003, where an amorphous aluminium fluoride phase pre- pared by sol–gel chemistry or other methods, such as
pared by a two-step fluorolytic sol–gel synthesis was devel- mechanochemistry.6–9 The general consensus is that if
oped.3 The benefit of the mesoporous HS-AlF3 phase obtained aqueous HF is used as a source for fluorination, or water is
that way is its high-surface area, as determined by BET experi- added to the reaction mixture, the final product will be a
ments, exhibiting extremely strong Lewis acidic surface sites. metal hydroxide fluoride. With respect to their acidic sites,
For comparison, the metastable β-AlF3 phase, classically the these solids possess both acid types, Lewis and Brønsted, and
most active AlF3 polymorph, has a surface area ten times lower their relative amount can be adjusted via the F/OH ratio that
exhibiting macroporosity and just medium strong Lewis strongly depends on the amount of water added.10,11
surface sites.2
Metal hydroxide fluorides obtained this way are, unless
thermally treated, X-ray amorphous materials, and possess
strong acid sites as determined by ammonia TPD.9,12 They also
show exciting catalytic properties. Thus, Agirrezabal-Telleria
et al. used a partially hydroxylated magnesium fluoride for the
dehydration of xylose to furfural, and Hemmann and Teinz
reported on the cyclization of citronellal to isopulegol with
high selectivity at different aluminium hydroxide fluoride
phases.13–15 Although aluminium hydroxide fluorides rep-
resent a very interesting new class of bi-acidic solid catalysts,
aDepartment of Chemistry, Humboldt-Universität zu Berlin, Brook-Taylor-Straße 2,
D-12489 Berlin, Germany. E-mail: thomas.braun@cms.hu-berlin.de,
erhard.kemnitz@chemie.hu-berlin.de
bSchool of Analytical Sciences Adlershof (SALSA), Humboldt-Universität zu Berlin,
Unter den Linden 6, 10099 Berlin, Germany
†Dedicated to Dr Johannes Eicher on the occasion of his 60th birthday.
‡Electronic supplementary information (ESI) available. See DOI: 10.1039/
c9dt00831d
This journal is © The Royal Society of Chemistry 2019
Dalton Trans.

View Article Online
Paper
Dalton Transactions
duced in this work were prepared as depicted in Fig. 2, using a
similar procedure. Along the paper, the following nomencla-
ture will be used: AlF3-OH refers to the xerogel obtained by
drying the gel that has been formed after aqueous HF addition
to aluminium isopropoxide; AlF3-OHpost refers to the solid
obtained after postfluorination of AlF3-OH. Nb-AlF3-OH and
Nb-AlF3-OHpost are the corresponding doped variants, pre-
pared with the addition of 5% mol of Nb to Al.
Fig. 1 Pictorial description of the catalysts surface.
Aluminium isopropoxide (Sigma Aldrich, 98%) was dis-
solved in dry isopropanol –previously distilled under mole-
cular sieves– and heated under reflux for 1 hour. After cooling
down the mixture, a solution of aqueous HF (39.32 M, 78%)
was added, in stoichiometric conditions. A transparent gel was
formed, which was left to age for 24 hours before removal of
the solvent under vacuum. The xerogel obtained was then
stored in a glovebox (AlF3-OH). For the preparation of AlF3-
OHpost, AlF3-OH was treated further in a post-fluorination step.
For this, the sample was treated at 240 °C under a flow of N2
and CHClF2 (ratio of 4 : 1) in a Ni reactor. In the case of the
Nb-doped sample, the desired amount (5% mol of Nb to Al) of
niobium ethoxide (Strem Chemicals, 99.9%) was added with a
syringe to the aluminium isopropoxide in isopropanol and left
to dissolve. Aqueous HF was added stoichiometrically for full
fluorination of both metal sources and the solvent was
removed to obtain a xerogel (Nb-AlF3-OH). A post-fluorination
treatment with CHClF2 was also performed, in this case the
full catalytic Lewis activity was achieved already at 200 °C, pro-
no AlF3-phase nor any other metal fluoride phase has been
reported so far that exhibits very strong Lewis surface sites
when Brønsted sites co-exist in a near neighbourhood. It is
assumed that Al–OH surface sites cause the formation of
Al–O–H⋯F–Al hydrogen bridges, resulting in a rearrangement
at the surface, which in turn blocks the very strong Al-Lewis
sites on the surface (see Fig. 1).16
We recently reported on the preparation of Nb-doped HS-
AlF3 by reacting alkoxide precursors with anhydrous HF in iso-
propanol. The presence of Nb unfortunately caused a decrease
in the Lewis acidity compared to the undoped material, as
indicated by the catalytic performance of the materials, however
Brønsted acid sites were also observed at the catalyst.17 The
activity in the isomerization of 1,2-dibromohexafluoropropane,
a reaction that only occurs at very strong Lewis acids like SbF5,
aluminium chlorofluoride, or HS-AlF3, was significantly lower
for Nb-doped HS-AlF3 than undoped HS-AlF3 (15 to 99%).
Based on these results, in this paper we present a new syn-
thetic approach by using HFaq to access a Nb-doped HS-AlF3
that contains Brønsted-sites besides very strong Lewis sites and
is able to perform the above-mentioned isomerization reaction
of 1,2-dibromohexafluoropropane with nearly 100% conversion.
In this approach, we make use of the fact that the thermo-
dynamic stability of the Nb–O bond in comparison to the Nb–F
bond is significantly more pronounced than the Al–O to Al–F
bond. The preparation involves the intermediate formation of
aluminium hydroxide fluoride phases as precursors, by using
aqueous HF as fluorinating agent instead of alcoholic solutions
of anhydrous HF. The aluminium hydroxide fluoride phase and
the Nb-doped variation were both post-fluorinated; all four
materials were thoroughly characterized, in particular by MAS
NMR spectroscopy and tested in specific catalytic reactions.
ducing the catalyst Nb-AlF3-OHpost
.
For comparison along this report, HS-AlF3 and Nb5
(HS-AlF3 with 5% mol of Nb – in this report herein named as
Nb-AlF3-post) were prepared following reported procedures.3,17
The main difference between these two catalysts (HS-AlF3 and
Nb5) and the ones presented here is the HF source used for
fluorination: anhydrous HF dissolved in isopropanol (HFiPrOH
)
instead of aqueous HF (HFaq, 39.32 M, 78%) which was used
here.
NH3-TPD
To determine the number and strength of the acid sites, the
catalysts were characterized by temperature programmed de-
sorption (TPD) using ammonia as a probe molecule. For this,
0.2 g of the catalyst were added in a quartz flow reactor and
heated at 300 °C under nitrogen flow for 1 hour. After cooling
down until 120 °C, ammonia was adsorbed on the surface, and
the excess removed by nitrogen flow for 1 h. The desorption
was performed using a heating rate of 10 °C min−1 up to
Experimental details
Catalyst preparation
All the reactions, unless otherwise stated, were performed 500 °C, kept at 500 °C until no more NH3 is desorbed. The
under standard Schlenk conditions. The four catalysts intro- process was monitored by FT-IR spectroscopy (FT-IR System,
Fig. 2 Scheme of catalysts preparation, with the nomenclature used in this work.
Dalton Trans.
This journal is © The Royal Society of Chemistry 2019

View Article Online
Dalton Transactions
Paper
PerkinElmer) following the band at 930 cm−1. The total the glovebox and covered with Kapton® film during the whole
amount of ammonia desorbed was quantified by back-titration measurement.
with a sodium hydroxide solution after treatment with excess
Surface area determination by N2 adsorption
sulfuric acid.
Surface area (using the BET model) and porosity (assessed
with the BJH model) were measured using nitrogen sorption at
−196 °C. An ASAP 2010 (Micromeritics Instrument Corp., USA)
instrument was used, and the samples were degassed for 6 h
either at 50 °C (for the precursors) or at 200 °C (for the post-
fluorinated solids).
Pyridine-FT-IR-PAS
To identify the presence of Lewis and/or Brønsted acid sites in
the catalyst, pyridine was used as a probe molecule in a FT-IR
photoacoustic spectroscopy (PAS) experiment. For this, the
sample was loaded with pyridine at 150 °C and left under N2
flow to remove physisorbed species. Afterwards, it was trans-
ferred into the photoacoustic cell (MTEC 200) and the respect-
ive IR spectrum was recorded at room temperature using a
FTIR system 2000, PerkinElmer. Background spectra, without
loading of probe molecule, were additionally measured for
each sample.
Dismutation of CHClF2 (Fig. 3a)
This test reaction can be performed in situ during the post-
fluorination step and followed at different temperatures.3,19 At
room temperature, only strong Lewis acids can show full con-
version. The flow reactor used for the final synthetic step was
built with a septum in the outlet from which samples could be
taken and measured. A GC-2010 (Shimadzu) equipped with an
HP-1 capillary column and an FID detector, with nitrogen as
carrier gas, was used to follow the reaction. Conversions were
calculated relatively from the ratio of CHClF2 to the different
products (CHF3, CHCl3, CHCl2F).
1H, 19F and 27Al MAS NMR spectroscopy
Solid-state MAS (magic angle spinning) nuclear magnetic reso-
nance spectra were recorded on a Bruker AVANCE 400 spectro-
meter at room temperature. The samples were filled in 2.5 mm
rotors in a glovebox to avoid contact with moisture. The
spectra were recorded using a rotation frequency of 20 kHz
1
(for 19F, H and 27Al) or 25 kHz (for 19F). All 19F spectra were
Isomerization of 1,2-dibromohexafluoropropane to 2,2-
dibromohexafluoropropane (Fig. 3b)
measured using a π/2 pulse length of 4.2 μs, a spectrum width
of 400 kHz, a recycle delay of 5 s and an accumulation number
of 32. All 1H spectra were measured using a π/2 pulse length of
3.8 μs, a spectrum width of 400 kHz, a recycle delay of 5 s and
an accumulation number of 256. Recycle delays longer than 5
s were tested to ensure the completeness of the signals and to
inspect their spin–lattice relaxation behaviour. Existent back-
ground signals were suppressed with the application of a
phase-cycled depth pulse sequence according to Cory and
To 25 mg of the catalyst, 250 μL of the substrate (Alfa Aesar,
98%) were added, and the mixture was left to react for 2 hours
at room temperature. Water was added to stop the reaction,
and CDCl3 to dissolve the product: the organic phase was sep-
arated and subjected to 19F NMR analysis. Conversions were
Ritchey.18 The rotor-synchronized 19F and H spin-echo experi-
1
ments were registered with a recycle delay of 5 s, an accumu-
1
lation number of 1024 for 19F and of 256 for H respectively,
and a dipolar evolution time of 0.5 ms. 27Al MAS NMR (I = 5/2)
spectra were recorded with an excitation pulse duration of
1
1 μs, and an accumulation number of 512. H chemical shifts
are referred to TMS, and adamantane was used as a secondary
standard for calibration. 19F chemical shifts are referred to δ =
0 ppm of CFCl3, 27Al chemical shifts are given with respect to
δ = 0 ppm of 1 M AlCl3 solution. For both nuclei, α-AlF3 was
used as a secondary standard for calibration.
Elemental analysis
Carbon, hydrogen and nitrogen content of the samples were
determined by elemental analysis by means of a EURO EA
device (HEKATech GmbH). The samples were prepared under
inert atmosphere during handling.
P-XRD
Fig. 3 Scheme of catalytic reactions tested: (a) dismutation of CHClF2,
(b) isomerization of 1,2-dibromohexafluoropropane and (c) cyclization
of citronellal to isopulegol (A), neo-isopulegol (B), iso-isopulegol (C) and
Powder-X-ray-diffractograms were recorded on an XRD-
3003-TT diffractometer (Seiffert & Co., Freiberg), with Cu-Kα as
radiation source (l = 1.542 Å). The samples were prepared in neoiso-isopulegol (D).
This journal is © The Royal Society of Chemistry 2019
Dalton Trans.

View Article Online
Paper
Dalton Transactions
calculated via integration of the signals of substrate and mation resulting in a blocking of very strong Lewis surface
product in the spectra.
sites.19,20 However, a higher degree of fluorination already
during the sol–gel synthesis can suppress the problem of coke
formation.
Cyclization of citronellal (Fig. 3c)
To prepare AlF3-OHpost and Nb-AlF3-OHpost, the corres-
ponding xerogels AlF3-OH and Nb-AlF3-OH were subjected to a
post-fluorination step. The combination of thermal treatment
and reaction with a fluorinating agent such as CHClF2 permits
the preparation of fully fluorinated solids. To probe the com-
pleteness of this postfluorination step, the full conversion of
the fluorinating agent in the dismutation reaction was used as
a positive control. These post-fluorinated catalysts AlF3-OHpost
and Nb-AlF3-OHpost are brown granular solids, and as their
xerogel counterparts, amorphous based on P-XRD.
Considering the carbon content, this is significantly reduced
in the solids after the postfluorination step and is as little as
2.6% and 0.9% for AlF3-OHpost and Nb-AlF3-OHpost, respect-
Inside a glovebox, a Teflon tube was filled with 25 mg of the
catalyst and sealed with a rubber septum. The tube was
inserted in an Eppendorf Thermomixer comfort, and toluene
(3 mL), citronellal (0.3 mL) and undecane (0.15 mL) as an
internal standard were added. The mixture was stirred 6 h at
80 °C at 600 rpm, then cooled down in an ice bath. The cata-
lyst was filtered off and the respective solution was investigated
by GC-FID; retention times were compared to those reported.15
Results and discussion
Catalyst synthesis
In this work, the original synthesis of HS-AlF3 was modified ively (see Table S1‡). However, since HS-AlF3 obtained from the
using an aqueous solution of HF instead of anhydrous HF to reaction of Al(OiPr)3 with a-HFiPrOH contains 1.6% carbon after
fluorinate aluminium isopropoxide, and simultaneously postfluorination, the use of HFaq. as a fluorinating source does
niobium ethoxide in the case of the Nb-doped HS-AlF3. Fig. 2 not have an apparently significant effect on the amount of the
depicts the synthesis pathways, which are analogous to the remaining organic content in the final post-fluorinated
ones already described.17
material.
Regarding the described synthesis routes, two comments
AlF3-OH and Nb-AlF3-AlOH were prepared by the addition
of aqueous HF to the solution of aluminium isopropoxide and should be pointed out. Firstly, just as the original route to Nb-
niobium ethoxide in isopropanol. A transparent gel was doped HS-AlF3, the inclusion of Nb centres generates enough
obtained, comparable to the ones obtained via the water-free Lewis acid sites in the catalyst to achieve full catalytic activity
route mentioned beforehand.17 After solvent removal, xerogels in the dismutation of CHClF2 during post-fluorination already
were obtained in both cases as fine white powders. To charac- at 200 °C. A higher temperature to enable full conversion of
terize them, elemental analysis and P-XRD were carried out: chlorodifluoromethane is not required, as in the case of AlF3-
both AlF3-OH and Nb-AlF3-AlOH are amorphous, and possess OHpost or HS-AlF3. Secondly, the effect of using aqueous HF
low carbon content (8.2% and 10.3% respectively, see obviously causes to some part not only a higher degree of
Table S1‡) when compared to phases obtained from reaction fluorination but also a higher extent of hydrolysis. This issue
with a-HFiPrOH. Since there are still organic species in the was further studied with several characterization methods
materials, according to elemental analysis, these two catalysts which are presented in the next sections.
should more correctly be considered as aluminium hydroxo-
alkoxo-fluorides (Al(OR)3−x(OH)yFz).
The Nb-AlF3-OHpost catalyst behaves very similar as HS-AlF3,
it is stable for a long time if stored in N2 atmosphere, but
If AlF3-OH and Nb-AlF3-OH are compared to the xerogel pre- takes up water if handled at air. Consequently, the very strong
cursor of HS-AlF3, a remarkable difference arises. Although all surface Lewis sites are blocked resulting in reduced catalytic
three compounds are X-ray amorphous, the carbon content for activity. However, as known for HS-AlF3, full catalytic activity
the HS-AlF3 xerogel prepared with a-HFiPrOH is with ca. 30% can be recovered by heating the Nb-AlF3-OHpost catalyst in a
significantly higher. Surprisingly, using aqueous HF solutions dry nitrogen flow up to ca. 300 °C (see ref. 20 and 21).
for fluorination resulted in drastically lower carbon contents
for AlF3-OH and Nb-AlF3-OH. It is evident from these elemen-
tal analysis results that aqueous HF causes a higher degree of
N2 adsorption experiments
conversion on the alkoxide, compared to a-HFiPrOH. The The adsorption of N2 at −196 °C provides crucial morphologi-
number of unreacted alkoxide groups in AlF3-OH and Nb-AlF3- cal information for any heterogeneous catalyst: surface area,
OH is lower, due to an extended dissociation of HF resulting pore size and pore volume. This set of information is summar-
possibly in increased reactivity.
ized in Table 1 for the different catalysts studied. The corres-
The lower amount of carbon in AlF3-OH and Nb-AlF3-OH ponding isotherms are depicted in Fig. 4, for both AlF3-OHpost
/
presents the first advantage over the sample obtained from the AlF3-OH (left) and Nb-AlF3-OHpost/Nb-AlF3-OH(right). All the
synthesis of HS-AlF3 using a-HFiPrOH. It is known, that in the catalysts exhibit high BET surface areas, the highest in particu-
course of postfluorination of the (Al(OR)3−x(OH)yFz) phases lar for both AlF3-OH and Nb-AlF3-OH (Table 1, entries 4 and
extremely strong Lewis sites are formed, which – if the post- 6), with values of ca. 460 m2 g−1. The effect of the postfluorina-
fluorination with a chlorofluorocarbon gas stream is not care- tion step results in a decrease of the surface area, and an
fully performed – can be accompanied by significant coke for- increase of both, pore size and pore volume, after the treat-
Dalton Trans.
This journal is © The Royal Society of Chemistry 2019

View Article Online
Dalton Transactions
Paper
Table 1 Summary of characterization
BET
Acid sites
per surface
area
surface Pore Pore
Acid
area
sizea volume sitesb
Entry Catalyst
[m2 g−1] [Å] [cm3 g−1] [μmol g−1] [μmol m−2
]
c
1
2
3
4
5
6
HS-AlF3
240
251
255
467
69 0.50
50 0.37
165 1.19
73 0.93
72 0.59
40 0.41
1510
1303
1968
1491
1843
1810
6.29
5.19
7.71
3.19
6.90
3.93
c
Nb-AlF3-post
AlF3-OHpost
AlF3-OH
Nb-AlF3-OHpost 267
Nb-AlF3-OH 460
a Average pore size, as calculated from BJH model (4
b Determined by NH3-TPD. c From ref. 17.
A
V−1).
Fig. 5 27Al MAS NMR spectra for AlF3-OH and AlF3-OHpost (left) and
Nb-AlF3-OH and Nb-AlF3-OHpost (right) measured at 20 kHz rotation
frequency, ns = 512; the chemical shifts of the signal maxima are indi-
cated in the figure.
ment at 200 °C (for Nb-AlF3-OHpost) or 240 °C (for AlF3-OHpost
with CHClF2.
The isotherms can be classified as type IV according to
)
MAS NMR spectroscopy
IUPAC, typical for mesoporous solids.22 Nb-AlF3-OH has a MAS NMR spectroscopy is an important analytical tool for
different shape compared to the others, due to a broader dis- these catalysts to gain information about local environments
tribution in pore sizes (see Fig. S1‡). All four isotherms show of such XRD amorphous solids. Aluminium hydroxide fluoride
differences with respect to adsorption and desorption and phases possess three NMR active nuclei which can be studied:
such hysteresis can be identified as type H3 and, in the case of 1H, 19F and 27Al. Several experiments were performed and the
Nb-AlF3-OH, type H4. They correspond to aggregates of par- obtained results are presented in Fig. 5 (27Al MAS NMR
ticles with slit-shaped pores, which have a non-uniform (for spectra), Fig. 6 (19F MAS NMR spectra), Fig. 7 (19F spin-echo-
H3) or uniform (for H4) shape and size.23
rotor-synchronized spectra) and Fig. 8 (1H spin-echo-rotor-syn-
Pore size distributions are shown in Fig. S1,‡ applying the chronized spectra and 1H MAS NMR spectra).
BJH model, for all four catalysts. The undoped ones, AlF3-
27Al MAS NMR spectroscopy. The 27Al MAS NMR spectra
OHpost and AlF3-OH, have quite similar distributions, with show a broad peak for all samples at chemical shifts varying
pore widths in the mesopore region. Particularly, AlF3-OH has from −11 to −16 ppm (Fig. 5). These signals are typical for alu-
also a micropore contribution, which under the experimental minium in a distorted octahedral F− environment and are also
conditions used here cannot be determined, and pore sizes commonly found for HS-AlF3. The asymmetric high-field decay
should be below 20 Å. The Nb-doped catalysts, on the other is related to the distribution of varying bond lengths and bond
hand, show different pore size distribution. The post-fluori- angles existing in the solid.
nated catalyst, Nb-AlF3-OHpost
whereas Nb-AlF3-OH possesses a mixture of mesopores and chemical shifts for several crystalline species of aluminium
micropores, as seen from the broad shape of the distribution.
(hydroxide) fluorides AlFx(OH)3−x, considering crystalline com-
,
exhibits only mesopores,
König et al.24 elaborated a trend analysis for 27Al NMR
Fig. 4 N2 adsorption and desorption isotherms at −196 °C for AlF3-OH and AlF3-OHpost (left) and Nb-AlF3-OH and Nb-AlF3-OHpost (right).
This journal is © The Royal Society of Chemistry 2019
Dalton Trans.

View Article Online
Paper
Dalton Transactions
pounds formed by different but AlFxO(x−y)/2 polyhedral units of
known structure.7,8 Amorphous compounds like aluminium
isopropoxide fluoride can be interpreted on using the same
analysis since they also consist of distorted octahedral
AlFx(OX)x−6-units (X being an H or alkyl moiety).25 Based on
these studies, a 27Al signal at about −11 ppm, as observed here
for AlF3-OH and Nb-AlF3-OH, can be assigned to aluminium
atoms in a coordination sphere of the type AlF5O, whereas the
value of −16 ppm present in the spectra of post-fluorinated
materials, can be assigned to AlF6 species. The origin of the
AlF5O species cannot be clearly assigned since they could
correspond either to hydroxide or isopropoxide moieties co-
ordinated to aluminium.25
19F MAS NMR spectroscopy. The 19F MAS spectra (Fig. 6)
exhibit only one broad signal, between −164 ppm and
−168 ppm, which corresponds to fluorine atoms in a bridging
position between two aluminium centres, and bonded to alu-
minium in an octahedral environment. Due to the shape and
widths of these signals, it can be assumed that many different
fluorine species are present. In the case of AlF3-OHpost and Nb-
AlF3-OHpost, a small signal was additionally observed at
−210 ppm, which is characteristic for terminal fluorine atoms
bound to sub-coordinated surface aluminium sites.26
Fig. 6 19F MAS NMR spectra for Nb-AlF3-OH (down-left) and Nb-AlF3-
OHpost (down-right), measured at 20 kHz rotation frequency, ns = 32
and for AlF3-OH (up-left, 20 kHz) and AlF3-OHpost (up-right), measured
at 25 kHz, ns = 32 (*: spinning side bands).
In addition, and to better comprehend these signals, spin-
echo-rotor-synchronized experiments were carried out. The
results for the catalysts studied here are presented in Fig. 7.
The rotor-synchronized spin-echo 19F MAS NMR spectrum of
the Nb-AlF3-OH phase depicts two signals, one at −150 ppm
and another one with a lower intensity at −172 ppm (Fig. 7,
left, bottom-down). Both signals are covered in the 19F MAS
NMR spectrum (Fig. 6), and can be assigned according to a 19
F
NMR chemical shift trend analysis to AlFxO6−x building blocks
as reported in the literature.27 It suggests that the smaller the
number of fluorine atoms at an aluminium centre is, the
stronger is the low-field shift of the 19F NMR signal. Therefore,
the signal at −150 ppm can be interpreted resulting from an
AlF4O2 unit, and the one at −172 ppm from an AlF6 species.
The spin-echo spectrum of AlF3-OH (Fig. 7, left-up) exhibits
two signals, at −176 ppm and at −196 ppm. The latter corres-
ponds to a terminal fluorine species, however, it is not intense.
The signal at −176 ppm can be assigned to an AlF6 unit and is
also present in the corresponding post-fluorinated catalyst.
Besides these two, an additional sharp signal appears at
−81 ppm: this signal corresponds to a –CF3 moiety, and might
belong to impurities from fluorinated grease used in the
synthesis.
Both spectra of AlF3-OHpost and Nb-AlF3-OHpost exhibit
sharp signals between −202 and −211 ppm, consistent with
the presence of terminal non-bridging fluorine atoms in these
catalysts, usually being a clear hint for very strong Lewis
acidity of the solid AlF3-phase.2 Thus, the addition of Nb into
the matrix does not affect to a great extent the presence of
such terminal sites. Furthermore, both, AlF3-OHpost and
Nb-AlF3-OHpost samples, show comparable 19F MAS NMR data
to their classical counterpart HS-AlF3, suggesting likewise a
high catalytic activity.
Fig. 7 19F MAS spin-echo-rotor-synchronized NMR spectra for Nb-
AlF3-OH (down-left) and Nb-AlF3-OHpost (down-right), measured at 20
kHz rotation frequency, a dipolar evolution time of 0.5 ms, ns = 1024;
and for AlF3-OH (up-left, 20 kHz) and AlF3-OHpost (up-right, νrot = 25
kHz), dipolar evolution time: 0.5 ms, ns = 1024.
Dalton Trans.
This journal is © The Royal Society of Chemistry 2019

View Article Online
Dalton Transactions
Paper
Fig. 8 (a) 1H MAS NMR spectra for AlF3-OH (up-left), AlF3-OHpost (up-right), Nb-AlF3-OH (down-left) and Nb-AlF3-OHpost (down-right). All spectra
were measured with a rotation frequency of 20 kHz, a recycle delay of 5 s, and an accumulation number of 256. (b) 1H MAS spin-echo-rotor-syn-
chronized NMR spectrum for Nb-AlF3-OH (down-left) and Nb-AlF3-OHpost (down-right) and for AlF3-OH (up-left) and AlF3-OHpost (up-right); νrot
20 kHz, dipolar evolution time: 0.5 ms, ns = 256, p1 = 3.8 μs.
=
Note, fluorine atoms generally do not exhibit different
For AlF3-OHpost and Nb-AlF3-OHpost, the spectra reveal
chemical shift when at an aluminium centre the neighbouring again similar signal patterns, indicating that the addition of
oxygen atom belongs to an hydroxide or to an isopropoxide Nb does not modify to a great extent the structure of AlF3-
moiety. Therefore both isopropoxide and hydroxide groups OHpost. Narrow signals appear at lower chemical shifts, 0.8 and
might be present in AlF3-OHpost and Nb-AlF3-OHpost since their 1.2 ppm, corresponding to methyl groups of the isopropoxide
signals are not distinguishable.
groups still unreacted. The signals at 2.8–5.3 ppm can be
1H MAS NMR spectroscopy. 1H MAS NMR spectroscopy assigned to the CH groups. Between 4 and 10 ppm, a broad
should provide information on hydrogen-containing moieties shoulder appears in both spectra, with slightly higher intensity
since according to elemental analysis, both AlF3-OH and at 7.5 ppm. Thus, at 7.5 ppm Nb-AlF3-OHpost exhibits a more
Nb-AlF3-OH exhibit less than 2.6% hydrogen content. At first intense signal than AlF3-OHpost, and this might be an indi-
glance, the signals (see Fig. 8A) are all quite broad and are cation for a larger number of bridging –OH species. Literature
hard to separate from each other. Even the high rotation fre- data for zeolites assign very strong OH bridging groups to a
quency used (20 kHz) is not enough to resolve such spectra, signal at 7.7 ppm and OH terminal groups bound at alu-
and obviously, a large number of different hydrogen contain- minium to a signal at 0.8 ppm.30
ing moieties are present in all the samples.
The broad shoulders at high chemical shift values,
The spectra of AlF3-OH and Nb-AlF3-OH exhibit a compar- characterized by peaks at 7.7 and 7.5 ppm, for AlF3-OHpost
able pattern with signals at 1.2 ppm, 4.3 ppm and 6.8 ppm, and Nb-AlF3-OHpost, respectively, show increased intensity
with different intensities. The chemical shifts and relative when compared to AlF3-OH and Nb-AlF3-OH. They arise
intensities of these signals can be assigned to –OiPr groups, from bridging OH groups and account for the Brønsted
and even adsorbed isopropanol based on studies by König.27,28 acidity that these catalysts possess. The data observed for
Bridging highly acidic–OH species, stemming from Al–OH these post-fluorinated phases can be understood by compar-
units, should be revealed by a signal at approximately 7.9 ppm ing it to previous work on fluoride-alkoxide phases.31 There,
according to Scholz et al.29 In the present case, the signals are the addition of small amounts of water in the course of the
too broad and they would cover such a signal; therefore, a fluorolytic sol–gel synthesis of HS-AlF3 increased the inten-
1
definite conclusion cannot be drawn. It is conceivable that sity of the shoulders between 4 and 10 ppm, too. A H MAS
AlF3-OH and Nb-AlF3-OH do possess bridging –OH groups, but NMR spectrum of an HS-AlF3 xerogel shows an intensity
the experimental conditions used to obtain the spectra do not ratio of 6 : 1 : 1 for the signals observed at 1.2 ppm, 4.3 ppm
allow to identify them unequivocally.
and 7 ppm; but this ratio changes with the addition of
This journal is © The Royal Society of Chemistry 2019
Dalton Trans.

View Article Online
Paper
Dalton Transactions
water, due to an increase in the number of bridging OH values are even higher than those of HS-AlF3, meaning that the
groups. fluorination with aqueous HF is a successful way to introduce
Fig. 8B depicts the 1H MAS spin-echo-rotor-synchronized a high density of acid sites in the porous matrix of the catalysts.
NMR spectra. Comparisons in intensity between different The profiles obtained during the thermal desorption of
solids are not possible anymore, since the environment of ammonia are depicted in Fig. 9. All the samples show a
each proton, and its relaxation, cannot be predicted or quanti- different acid strength distribution, from medium-strong (de-
fied. These experiments, however, confirm what has been sorption up to ca. 400 °C) to very strong acid sites (above
observed previously in the 1H MAS NMR spectra, and now 400 °C). In the case of AlF3-OH, the TPD exhibits a broad peak
signals for the different proton species are properly dis- from 150 °C to 420 °C, and a narrower one between 420 °C
tinguishable. The spectra of AlF3-OH and Nb-AlF3-OH are and 500 °C, both with similar intensities. The post-fluorinated
quite similar (Fig. 8B, left side). Signals at approximately 1, 4 variant, AlF3-OHpost, shows the same general behaviour, with
and 7 ppm can be identified, but overall they are still distinc- additional desorption peaks. In the case of the Nb-doped cata-
tively broad. Note that broad lines due to bridging OH species lysts, the two catalysts show different shape in the desorption.
show a lower intensity due to the dipolar evolution time in the Nb-AlF3-OH is more pronounced structured and shows four
echo experiment (cf. Fig. 8B and caption).
Many different signals appear in the echo experiment in strong acid sites, whereas Nb-AlF3-OHpost exhibits only two
the post-fluorinated phases AlF3-OHpost and Nb-AlF3-OHpost peaks, one characteristic for medium-strength sites and a
different desorption peaks, two of them corresponding to very
,
which before were not distinguishable under the broad smaller one for very strong sites at 500 °C. This suggests that
shoulder from 4 to 10 ppm. This is an indication that there are the postfluorination step for the Nb-doped catalyst does not
many different types of proton environments in the solid, produce an increase in the total number of acid sites (see
which again stresses the presence of several types of OH Table 1), but medium and strong sites seem to be converted
groups with different identities.
into weaker and very strong sites.
Pyridine-FT-IR-PAS. Pyridine coordinatively binds to Lewis
acid sites (Lpy) through the lone pair of electrons, whereas at
Brønsted acid sites the aromatic nitrogen of the ring can be
protonated (BpyH+).32–34 The vibrations of the pyridine ring
can be assigned to Lpy (1445 and 1490 cm−1) and to BpyH+
(1490 and 1545 cm−1) in the spectral window of 1400 to
1600 cm−1. If both types of acid sites are present in a solid
catalyst, the IR spectra will show three bands, two of which
can be assigned exclusively to Lpy (at 1445 cm−1) and to BpyH+
(at 1545 cm−1), and a third one which is an overlap of the
bands of both Lpy and BpyH+. Experimentally the method of
choice is photoacoustic spectroscopy, because powders or
opaque samples can be directly measured;32,35 moreover it
permits rapid measurements.
Acid sites characterization
NH3-TPD. To determine the overall number of acid sites and
the strength distribution, ammonia TPD experiments were
carried out. Table 1 shows the total number of acid sites and
data normalized by BET surface area for each catalyst studied,
as determined by quantification of the desorbed ammonia.
Remarkably, most catalysts exhibit an increased absolute
number of acid sites when compared to the HS-AlF3 or the
Nb-AlF3-post doped variant, with the exception of AlF3-OH,
which has a similar value. However, this is mainly due to the
evidently larger surface area but becomes negligible when
comparing the specific site density (number of acid sites per
m2). When comparing the precursor and post-fluorinated cata-
lysts, the postfluorination increases not only the total number
of acid sites but also the specific acid sites per surface area by
a factor of two due to the reduction in BET surface area. These
Fig. 10 shows the spectra obtained after pyridine adsorp-
tion. All four catalysts exhibit both kinds of acid sites, Lewis
and Brønsted, as indicated by the presence of three bands.
Fig. 9 NH3 TPD profiles for AlF3-OH and AlF3-OHpost (left) and Nb-AlF3-OH and Nb-AlF3-OHpost (right).
Dalton Trans.
This journal is © The Royal Society of Chemistry 2019

View Article Online
Dalton Transactions
Paper
Fig. 10 FT-IR-photoacoustic measurement after pyridine adsorption for AlF3-OH and AlF3-OHpost (left) and Nb-AlF3-OH and Nb-AlF3-OHpost
(right). Signals corresponding to Lewis acid site-bounded pyridine (LPy) and to Brønsted acid site-bounded pyridine (BPy).
Noticeably, the addition of Nb causes an increase of into very strong acid sites. Likewise, Al–OH groups in AlF3-OH
Brønsted acid sites, observed in the change of the ratio of Lpy are converted into strong Lewis acid sites (Al–F). There is no
to BpyH+ when compared to both AlF3-OHpost and AlF3-OH. remarkable difference in the NH3-TPD profile because
Quantification is not recommended due to the too low sensi- Brønsted acid sites are converted into Lewis acid sites of
tivity of this method, but apparently, Nb-AlF3-OHpost and Nb- apparently similar strength.
AlF3-OH possess more Brønsted than Lewis acid sites. AlF3-
As to the higher portion of Brønsted sites in all the Nb-AlF3-
OHpost and AlF3-OH exhibit a more intense signal for Lewis OHpost samples, a reasonable explanation is the fact that the
acid sites, however, Brønsted acid sites are still present but bonding energy of Nb–O (726 kJ mol−1) is higher than that of
minor. It is remarkable that the postfluorination step seems Nb–F bonds (452 kJ mol−1) whereas Al–O bonds (501 kJ mol−1
)
not to have a significant effect on the ratio of Brønsted vs. are significantly weaker than Al–F bonds (665 kJ mol−1).36
Lewis acid site distribution, and the higher intensities of the Thus, Nb–OH and/or Nb–OR groups that originate from the
peaks for AlF3-OHpost may indicate an increase of the overall synthetic approach are quite more difficult to postfluorinate as
number of acid sites in the sample. Even for Nb-AlF3-OHpost
the Brønsted acid sites increased just slightly.
,
compared to Al–OH and/or Al–OR groups. This also implies
that, once fluorinated, Nb–F bonds are less stable than Al–F
Taking together the NH3-TPD and the Py-FTIR results for bonds against the attack of water, and thus can undergo hydro-
AlF3-OHpost and AlF3-OH there seems no significant effect of lysis more easily.36 This data thus explain that the observed
the post fluorination step on the identity of acid sites higher amount of Brønsted sites in all the Nb samples is
(Brønsted vs. Lewis) and the site strength distribution. On the reasonable.
contrary, in the case of the niobium-containing catalysts, the
ratio of the type of acid sites seems also to be unchanged but
Catalytic reactions
the site strength distribution changed as a result of post-fluori-
Dismutation of CHClF2. One of the three reactions that were
nation. Moreover, considering the two peaks of the Nb-AlF3- chosen in order to probe the strength and kind of acid surface
OH sample at 500 °C and above 500 °C in the NH3-TPD sites (Lewis vs. Brønsted) of these new catalysts is the dismuta-
measurements, it implies that the post-fluorinated sample Nb- tion of chlorodifluoromethane, HCFC-22 (see Fig. 3a).
AlF3-OHpost even lost some very strong acid sites. The observed HCFC-22 is a greenhouse gas long time used as a refrigerant
differences between both pairs of catalysts (before and after in the past but still the intermediate for polytetrafluoroethyl-
post-fluorination) are not entirely clear. First of all, perhaps ene (PTFE) production and is an excellent smooth fluorinating
they are due to the presence of both Nb–F and Nb–OH species agent for preparing the post-fluorinated catalysts. An advan-
on the surface of the catalyst. The latter species, which orig- tageous side-effect is that during the course of the postfluori-
inate from the aqueous media used to fluorinate the alkoxides, nation the degree of catalytic dismutation of HCFC-22 at
is responsible for the Brønsted acidity, whereas the Nb–F and different temperatures can be used to monitor the formation
the majority of the Al–F sites in the solid account for Lewis of catalytic active Lewis sites. After two hours heterogeneous
acid sites. The doping with Nb centres to the AlF3-catalyst reaction between the catalyst precursor and HCFC-22 full cata-
obviously generates a larger number of terminal OH species as lytic dismutation activity had been achieved at already 200 °C
compared to the undoped one. After post-fluorination, most of for the formation of Nb-AlF3-OHpost and at 240 °C for AlF3-
those terminal Al–OH species in Nb-AlF3-OH transfer into Al–F OHpost. Then the reactor was cooled down and the dismutation
units and thus in Nb-AlF3-OHpost strong sites are converted conversion measured again at room temperature: this value is
This journal is © The Royal Society of Chemistry 2019
Dalton Trans.

View Article Online
Paper
Dalton Transactions
in itself an excellent indication of the presence of strong Lewis order to probe the extremely strong Lewis acidity of aluminium
acid sites.
chlorofluoride.37 Even with SbF5, considered classically as the
Table 2 illustrates the values obtained for these two post- strongest Lewis acid at all – now surpassed by various mole-
fluorinated catalysts, along with those for HS-AlF3 and Nb- cular Lewis superacids,38–40 only at 80 °C and 2 h reaction time
AlF3-post, which were in the course of the fluorolytic sol–gel syn- conversion above 90% is achieved. This reaction occurs at
thesis fluorinated with a-HF in isopropanol. All catalysts room temperature only in the presence of very strong Lewis
exhibit outstanding catalytic performance at room tempera- acid sites and can be considered as a benchmark reaction.2
ture, making AlF3-OHpost and Nb-AlF3-OHpost comparable in With AlCl3 no reaction takes place at room temperature and
activity with HS-AlF3. The Nb-containing catalyst already shows only after 3 days at 150 °C, 74% conversion is observed.41
full conversion at 200 °C, indicating an advanced creation of Therefore, reactivity at room temperature was tested (see
strong Lewis surface sites in the doped catalyst when com- Fig. 3b and Table 3). In the case of AlF3-OH and Nb-AlF3-OH,
pared with the un-doped HS-AlF3 catalyst, just as in the case of no reactivity was observed. On the other hand, both AlF3-
HS-AlF3 when compared with Nb-AlF3-post
.
OHpost and Nb-AlF3-OHpost turned out to be excellent catalysts,
Isomerization of 1,2-dibromohexafluoropropane. Isomerization exhibiting more than 90% conversion of 1,2-dibromohexafluor-
of dibromopolyfluoroalkanes was tested by Petrov et al. in opropane to 2,2-dibromohexafluoropropane after only 2 hours
at room temperature. These results confirm the observation
that the postfluorination step is crucial in creating very strong
acid sites by transforming the aluminium alkoxide groups that
Table 2 Dismutation of chlorodifluoromethane during post fluorination
still remain inside the AlF3-phases after their fluorolytic sol–
step
gel preparation into more reactive aluminium fluoride ones.
Remarkably, it is the first time that a AlF3 phase with a signifi-
cant content of Brønsted acidic surface sites exhibits very
strong Lewis sites that fully catalyse the isomerization of 1,2-
dibromohexafluoropropane.
Conversion
Conversion
Conversion
Entry Catalyst
at 200 °C a [%] at 240 °C a [%] at 30 °C a [%]
c
1
2
3
4
HS-AlF3
Nb-AlF3-post
AlF3-OHpost
2
97
1
96
99
99
98
99
c
b
99
For comparison, the above-mentioned catalysts, prepared
via fluorination with anhydrous HF in isopropanol, are also
included in Table 3. Nb-AlF3-post shows less than 20% conver-
sion in this particular reaction.17 This is in agreement with the
conclusions made at that time, since for the Nb-doped
samples a direct relationship between the decrease of reactivity
and higher amount of Nb centres in the catalyst samples was
observed. However, the aqueous synthesis route obviously
alters the material in such a way that the AlF3-based catalysts
obtained – even doped by Nb – display very strong Lewis acid
sites, thus affording full conversion for this reaction at room
b
Nb-AlF3-OHpost 97
a Calculated relatively by integration of peak area in its respective chro-
matogram. b Reaction performed until 200 °C. c From ref. 17.
Table 3 Isomerization of 1,2-dibromohexafluoropropane to 2,2-
dibromohexafluoropropanea
Entry
Catalyst
Conversion [%]
b
1
2
3
4
5
6
HS-AlF3
99
15
99
0
95
0
b
Nb-AlF3-post
AlF3-OHpost
AlF3-OH
temperature.
A plausible explanation for this difference
between Nb-AlF3−post and Nb-AlF3-OHpost originates from the
application of HFaq, which leads to a higher degree of fluorina-
tion during the sol–gel approach, due to the higher acidity.
Cyclization of citronellal. Pyridine FTIR spectroscopy clearly
evidenced the presence of both, Lewis and Brønsted acid sites,
Nb-AlF3-OHpost
Nb-AlF3-OH
a Reaction conditions: 2 h, 25 °C. Conversion calculated from inte-
gration of 19F NMR. b See ref. 17.
Table 4 Cyclization of citronellal to isopulegola
Conversion of citronellal
[%]
Total yield to isopulegolsb
[%]
Selectivity to isopulegol (A)c
[%]
Yield of isopulegol (A)
[%]
Entry
Catalyst
d
1
2
3
4
5
6
HS-AlF3
91.7
98.8
83.5
95.4
99.4
99.4
55.8
74.8
40.6
48.1
58.0
46.9
51.2
48.7
50.6
52.2
49.6
50.7
28.6
36.4
20.6
25.1
28.8
23.8
d
Nb-AlF3-post
AlF3-OHpost
AlF3-OH
Nb-AlF3-OHpost
Nb-AlF3-OH
a Reaction conditions: 6 h, 80 °C, 600 rpm; 3 mL toluene, 0.3 mL citronellal and 0.15 mL undecane as internal standard. All the values were calcu-
lated via GC/FID using a calibration curve for citronellal and the retention times were compared to the literature.15 A blank reaction without a
catalyst was carried out. Conversion of citronellal was quantified (4.7%) and the given values were corrected accordingly. b Calculated based on
the initial amount of citronellal, for all four possible isomers. c Selectivity to isopulegol (A) over the other three isomers. d See ref. 17.
Dalton Trans.
This journal is © The Royal Society of Chemistry 2019

View Article Online
Dalton Transactions
Paper
especially in the Nb-doped HS-AlF3. An interesting reaction to mutation of CHClF2 and isomerization of 1,2-dibromohexa-
test the bi-acidic character of catalysts is the cyclization of fluoropropane. A remarkable difference from the already
citronellal. This reaction is considered to proceed only in the known HS-AlF3 is that the use of HFaq permits to obtain a bi-
presence of a catalyst bearing both Lewis and Brønsted acid acidic catalyst, where very strong Lewis acid sites coexist with
sites and has been studied as a model acid-catalyzed reaction Brønsted acid sites. The Nb-doped aluminium fluoride system
by Chuah et al., since it is also of interest for the industrial presented here possess very strong Lewis acid sites, alongside
production of menthol.42 For a successful conversion and Nb-OH groups accounting for Brønsted acid sites. Remarkably,
selectivity to obtain the desired product isopulegol and their the presence of Nb allows the introduction of OH groups
corresponding isomers, two steps need to take place: the iso- which are preserved even after the post-fluorination treatment
merization of the double bond followed by protonation of the and do not affect the high strength of the Lewis acidic Al sites.
oxygen of the ketone group and ring closure. This mechanism
requires not only the presence of both Lewis and Brønsted
acid sites in the catalyst but also a distribution of those sites Conflicts of interest
in a neighbouring position.42,43
There are no conflicts to declare.
Consequently, the cyclization of citronellal (depicted in
Fig. 3c) was used to probe the presence of both kinds of acid
sites and to test all the four catalysts studied here. The results
are summarized in Table 4. Additional values are included for
Acknowledgements
comparison (HS-AlF3 and Nb-AlF3-post). All catalysts show excel-
lent conversions for citronellal (more than 83%), with a slight
Dr Andrea Zehl, for elemental analysis, and Ms. Sigrid Bäßler,
for FT-IR-PAS and TPD measurements, are gratefully
improvement for the Nb-doped ones (up to 99%). The total
acknowledged. C. P. M. would like to thank Dr Cecilia
yield for all the isopulegol isomers is higher than 40%, and in
Spedalieri and Dr Patricia Russo for fruitful discussions. The
the case of Nb-AlF3-OHpost, the best result can be obtained,
authors would also like to thank the financial support from
with a 58% yield. The selectivity to isopulegol (A) over the
the graduate school SALSA (School of Analytical Sciences
Adlershof) and the CRC 1349 (project number 387284271),
both funded by the Deutsche Forschungsgemeinschaft (DFG,
German Research Foundation).
other three isomers is similar for all the catalysts.
The addition of Nb centres seems to be a factor that
enhances conversion and total yield, and its effect is more pro-
nounced when the catalysts are post-fluorinated (e.g. compar-
ing AlF3-OHpost vs. Nb-AlF3-OHpost). Nb-AlF3-OHpost shows the
best performance, with the highest conversion of citronellal
and a very good overall yield for all isopulegol isomers. For
References
AlF3-OH and AlF3-OHpost, the postfluorination step leads to a
worsening of the catalytic performance. Two hypotheses can
be provided in order to rationalize this effect. A first possible
reason might be that the Brønsted acid sites which are needed
for this reaction have been drastically reduced in the course of
post-fluorination in AlF3-OHpost. Due to the strong difference
in the binding energies between Al–OH and Al–F bonds, the
majority of Al–OH bonds (which account for the Brønsted
acidity) are converted into Al–F bonds (which account for the
Lewis acidity). Thus, conversion and selectivity towards isopu-
legols decrease. The situation is different in case of the Nb-
doped catalysts. Nb-AlF3-OHpost and Nb-AlF3-OH, contrary to
their undoped variants, do still possess a higher proportion of
Brønsted acid sites originating from Nb–OH surface sites.
Additionally, it seems that OH-groups bound at Nb-sites do
not interfere that much with Lewis Al-surface sites as Al–OH
groups at the same Al-sites do.
1 G. Busca, Heterogeneous Catalytic Materials: Solid State
Chemistry, Surface Chemistry and Catalytic Behaviour,
Elsevier B.V., 2014.
2 T. Krahl and E. Kemnitz, Catal. Sci. Technol., 2017, 7, 773–
796.
3 E. Kemnitz, U. Groß, S. Rüdiger and C. S. Shekar, Angew.
Chem., Int. Ed., 2003, 42, 4251–4254.
4 S. Rüdiger and E. Kemnitz, Dalton Trans., 2008, 9226, 1117.
5 E. Kemnitz, Catal. Sci. Technol., 2015, 5, 786–806.
6 G. Scholz, S. Brehme, M. Balski, R. König and E. Kemnitz,
Solid State Sci., 2010, 12, 1500–1506.
7 D. Dambournet, A. Demourgues, C. Martineau, S. Pechev,
J. Lhoste, J. Majimel, A. Vimont, J. C. Lavalley, C. Legein,
J. Y. Buzaré, F. Fayon and A. Tressaud, Chem. Mater., 2008,
20, 1459–1469.
8 D. Dambournet, A. Demourgues, C. Martineau, E. Durand,
J. Majimel, A. Vimont, H. Leclerc, J.-C. Lavalley, M. Daturi,
C. Legein, J.-Y. Buzaré, F. Fayon and A. Tressaud, J. Mater.
Chem., 2008, 18, 2483–2492.
9 C. Stosiek, G. Scholz, G. Eltanany, R. Bertram and
E. Kemnitz, Chem. Mater., 2008, 20, 5687–5697.
Conclusion
In this paper we present a new, alternative and effective 10 F. Hemmann, G. Scholz, K. Scheurell, E. Kemnitz and
approach to prepare highly acidic aluminium fluoride phases. C. Jaeger, J. Phys. Chem. C, 2012, 116, 10580–10585.
Such AlF3-phases, after the postfluorination treatment, exhibit 11 F. Hemmann, I. Agirrezabal-Telleria, C. Jaeger and
an extremely high catalytic activity in reactions such as the dis-
E. Kemnitz, RSC Adv., 2015, 5, 89659–89668.
This journal is © The Royal Society of Chemistry 2019
Dalton Trans.

View Article Online
Paper
Dalton Transactions
12 C. Stosiek, G. Scholz, S. L. M. Schroeder and E. Kemnitz, 28 R. König, G. Scholz and E. Kemnitz, J. Phys. Chem. C, 2009,
Chem. Mater., 2010, 22, 2347–2356. 113, 6426–6438.
13 I. Agirrezabal-Telleria, F. Hemmann, C. Jäger, P. L. Arias 29 G. Scholz, S. Brehme, R. König, D. Heidemann and
and E. Kemnitz, J. Catal., 2013, 305, 81–91. E. Kemnitz, J. Phys. Chem. C, 2010, 114, 10535–10543.
14 F. Hemmann, C. Jaeger and E. Kemnitz, RSC Adv., 2014, 4, 30 Y. Jiang, J. Huang, W. Dai and M. Hunger, Solid State Nucl.
56900–56909.
Magn. Reson., 2011, 39, 116–141.
15 K. Teinz, PhD thesis, Humboldt Universität zu Berlin, 2012.
16 C. L. Bailey, S. Mukhopadhyay, A. Wander, B. G. Searle and
N. M. Harrison, J. Phys. Chem. C, 2009, 113, 4976–4983.
31 R. König, G. Scholz and E. Kemnitz, J. Sol-Gel Sci. Technol.,
2010, 56, 145–156.
32 J. Ryczkowski, Catal. Today, 2001, 68, 263–381.
17 C. P. Marshall, T. Braun and E. Kemnitz, Catal. Sci. 33 M. Tamura, K. Shimizu and A. Satsuma, Appl. Catal., A,
Technol., 2018, 8, 3151–3159. 2012, 433–434, 135–145.
18 D. G. Cory and W. M. Ritchey, J. Magn. Reson., 1988, 80, 34 M. C. Kung and H. H. Kung, Catal. Rev., 1985, 27, 425–
128–132. 460.
19 S. Rüdiger, G. Eltanany, U. Groß and E. Kemnitz, J. Sol-Gel 35 A. Rosencwaig and A. Gersho, J. Appl. Phys., 1976, 47, 64–
Sci. Technol., 2007, 41, 299–311. 69.
20 S. K. Rüdiger, U. Groß, M. Feist, H. A. Prescott, 36 Y.-R. Luo, Comprehensive Handbook of Chemical Bond
S. C. Shekar, S. I. Troyanov and E. Kemnitz, J. Mater. Chem.,
2005, 15, 588.
Energies, CRC Press Inc - M.U.A., 1st edn, 2007.
37 V. A. Petrov, C. G. Krespan and B. E. Smart, J. Fluor. Chem.,
1998, 89, 125–130.
38 L. O. Müller, D. Himmel, J. Stauffer, G. Steinfeld,
J. Slattery, G. Santiso-Quinones, V. Brecht and I. Krossing,
Angew. Chem., Int. Ed., 2008, 7659–7663.
39 J. F. Kögel, D. A. Sorokin, A. Khvorost, M. Scott, K. Harms,
D. Himmel, I. Krossing and J. Sundermeyer, Chem. Sci.,
2018, 9, 245–253.
21 T. Krahl, A. Vimont, G. Eltanany, M. Daturi and
E. Kemnitz, J. Phys. Chem. C, 2007, 111, 18317–18325.
22 M. Thommes, K. Kaneko, A. V. Neimark, J. P. Olivier,
F. Rodriguez-Reinoso, J. Rouquerol and K. S. W. Sing, Pure
Appl. Chem., 2015, 87, 1051–1069.
23 G. Leofanti, M. Padovan, G. Tozzola and B. Venturelli,
Catal. Today, 1998, 41, 207–219.
24 R. König, G. Scholz, A. Pawlik, C. Jäger, B. Van Rossum, 40 A. Wiesner, T. W. Gries, S. Steinhauer, H. Beckers and
H. Oschkinat and E. Kemnitz, J. Phys. Chem. C, 2008, 112,
15708–15720.
S. Riedel, Angew. Chem., Int. Ed., 2017, 56, 8263–
8266.
25 R. König, G. Scholz, A. Pawlik, C. Jager, B. Van Rossum and 41 P. A. Morken and D. J. Burton, J. Org. Chem., 1993, 58,
E. Kemnitz, J. Phys. Chem. C, 2009, 113, 15576–15585. 1167–1172.
26 T. Krahl, R. Stösser, E. Kemnitz, G. Scholz, M. Feist, G. Silly 42 G. K. Chuah, S. H. Liu, S. Jaenicke and L. J. Harrison,
and J.-Y. Buzaré, Inorg. Chem., 2003, 42, 6474–6483. J. Catal., 2001, 200, 352–359.
27 R. König, G. Scholz, R. Bertram and E. Kemnitz, J. Fluor. 43 Z. Yongzhong, N. Yuntong, S. Jaenicke and G. K. Chuah,
Chem., 2008, 129, 598–606. J. Catal., 2005, 229, 404–413.
Dalton Trans.
This journal is © The Royal Society of Chemistry 2019