Zeolite structure direction by simple bis(methylimidazolium) cations: The effect of the spacer length on structure direction and of the imidazolium ring orientation on the 19F NMR resonances

Article abstract of DOI:10.1021/ja210703y

A series of doubly charged structure-directing agents based on two methylimidazolium moieties linked by a linear bridge of n = 3,4,5, or 6 methylene groups has been used in the synthesis of pure silica zeolites in the presence of fluoride. All of them yielded zeolite TON while only the one with n = 4 was able to produce also zeolite MFI at highly concentrated conditions. In this MFI zeolite, two distinct ¹⁹F MAS NMR resonances with about equal intensity were observed, indicating two different chemical environments for occluded fluoride. With the singly charged 1-ethyl-3-methylimidazolium cation, which can be formally considered as the monomer of the bis-imidazolium cation with n = 4, TON and MFI were also obtained, and again two ¹⁹F MAS NMR resonances now with largely dissimilar intensities were observed in MFI. Molecular mechanics simulations support a commensurate structure-direction effect for n = 4 in MFI, with each imidazolium ring, in two different orientations, sitting close to the [4¹5²6 ²] cage. Periodic DFT calculations suggest that F in MFI resides always in the [4¹5²6²] cages, with the different ¹⁹F resonances observed being due to the different orientation of the closest imidazolium ring.

Full text of DOI:10.1021/ja210703y

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Article
pubs.acs.org/JACS
Zeolite Structure Direction by Simple Bis(methylimidazolium)
Cations: The Effect of the Spacer Length on Structure Direction and
19
of the Imidazolium Ring Orientation on the F NMR Resonances
Alex Rojas,† Luis Gomez-Hortiguela,‡ and Miguel A. Camblor*,†
́
̈
†Instituto de Ciencia de Materiales de Madrid (ICMM-CSIC), Sor Juana Ines
‡Instituto de Catal
isis y Petroleoquímica (ICP-CSIC), Marie Curie 2, Cantoblanco 28049, Madrid, Spain
* Supporting Information
́
de la Cruz 3, Cantoblanco 28049, Madrid, Spain
́
S
ABSTRACT: A series of doubly charged structure-directing agents based on
two methylimidazolium moieties linked by a linear bridge of n = 3,4,5, or 6
methylene groups has been used in the synthesis of pure silica zeolites in the
presence of fluoride. All of them yielded zeolite TON while only the one with
n = 4 was able to produce also zeolite MFI at highly concentrated conditions.
In this MFI zeolite, two distinct 19F MAS NMR resonances with about equal
intensity were observed, indicating two different chemical environments for
occluded fluoride. With the singly charged 1-ethyl-3-methylimidazolium
cation, which can be formally considered as the “monomer” of the bis-imidazolium cation with n = 4, TON and MFI were also
obtained, and again two 19F MAS NMR resonances now with largely dissimilar intensities were observed in MFI. Molecular
mechanics simulations support a commensurate structure-direction effect for n = 4 in MFI, with each imidazolium ring, in two
different orientations, sitting close to the [415262] cage. Periodic DFT calculations suggest that F in MFI resides always in the
[415262] cages, with the different 19F resonances observed being due to the different orientation of the closest imidazolium ring.
formation of EUO for n = 5, 69,10 and NEI for n = 10.11 This
INTRODUCTION
■
Structure direction1,2 by organic species plays a central role in
the development of new zeolite structures and compositions. In
addition to other effects typically intervening in the
crystallization of zeolites, the organic structure-directing agent
(SDA) usually ends up filling the zeolite voids providing a
stabilization energy that may help the zeolite nucleate and
grow. The inorganic−organic host−guest matching may range
from a loose pore filling, leading to unspecific structure
direction, to a close geometric fitting, providing a more robust
and specific effect.2,3 Despite the frequent cationic nature of
organic SDAs, electrostatic interactions are generally consid-
ered to contribute to a lesser extent in stabilizing the zeolite
frameworks than long-range van der Waals contacts.1,4
However, recent studies on a pure silica zeolite with the ITW
Framework Type Code5 containing 1,3,4-trimethylimidazolium
cations and fluoride anions supported that, in that case, a
significant stabilization of the pure silica framework is provided
by host−guest charge transfer, in addition to van der Waals
interactions.6,7
prompted Moini and co-workers to investigate the effect of
systematically varying the length of the spacers in this series of
diquats.12 Under particular synthesis conditions, the diquats
with n = 5−14 tended to yield zeolites with one-dimensional
pores (MTT, for n = 7, 8, 11, 12 or MTW, n = 9, 14), except
for specific values of n that gave two-dimensional or quasi-two-
dimensional pore zeolites: EUO (n = 5, 6) and NES (n = 10).12
The authors proposed that this specificity was related to the
correlation between the pore system of the two-dimensional
pore zeolites and the length of the corresponding diquats.12
These and other examples using doubly charged SDAs13,14
refer mainly to synthesis in hydroxide media, and the ra-
tionalization of the results are generally based more on space
filling considerations than on the effect of cationic charges.
This is so because, for aluminosilicates and silicates prepared
in hydroxide media, van der Waals interactions may be more
important than Coulombic interactions.1 For these materials,
framework anionic charges (either Al, Ga, or connectivity
defects), needed for charge balance, may have a range of pos-
sible locations (i.e., they generally do not show a very large
preference for sitting at specific positions). However, in pure
silica zeolites prepared in fluoride media, the cationic charge
of the SDA is balanced by occluded fluoride and this anion
has been shown to have a relatively large preference for sit-
ting in small cavities.15 Among those, double four ring cages
A number of studies have reported on the use as SDA of
multicharged organic species containing somewhat bulky
cationic moieties linked by flexible bridges of varying length
acting as spacers. This strategy was pioneered by Daniels and
co-workers,8 who used quaternary ammonium polymers based
on the 1,4-diazabicyclo[2.2.2]octane with linear alkane bridges
of varying length as spacers. Later on, several works focused on
polymethylene-bis(trimethylammonium) diquaternary com-
pounds, “diquats”, (CH3)3N+(CH2)nN+(CH3)3, reporting the
Received: November 14, 2011
Published: January 24, 2012
© 2012 American Chemical Society
3845
dx.doi.org/10.1021/ja210703y | J. Am. Chem. Soc. 2012, 134, 3845−3856

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Journal of the American Chemical Society
Article
(denoted as D4R or [46] cages) seem to be preferred,16,17 while
larger cavities containing some four rings are also frequent. Pre-
ferential location of fluoride in small cavities must be accom-
panied by sitting of the cation in larger spaces in its close
vicinity and it may even impose ordering on the organic
cations.18,19 Thus, we hypothesized that cation−anion pairings
between positive moieties of doubly charged cations and fluo-
ride occluded in small cavities of the silica wall could give rise to
specific structure-direction effects when the charges are
separated by a length commensurate to the distance between
F-containing cages. This has been studied before for poly-
methylene-bis(quinuclidinium),20 polymethylene-bis(trimethyl-
ammonium),21 and a series of diquats with different hetero-
cyclic cationic groups,22 but the discovery of charge transfer effects
in ITW prompted us to perform a study with polymethylene-
bis(methylimidazolium) molecules.
At different time intervals, the autoclaves were taken out from the
oven and quenched with tap water, and the solid products were
recovered by filtration and exhaustively washed with water. The final
composition of the reaction mixtures was SiO2:0.5 ROH:0.5 HF:w
H2O, and the H2O/SiO2 ratio was varied between 8.5 and 3.5. In the
above composition, R stands for an equivalent of the SDA used.
Characterization. The recovered solids were identified by powder
X-ray diffraction (XRD), recorded in a Bruker D8 Advance diffracto-
meter using Cu Kα radiation between 5 and 45° 2θ values. Multi-
nuclear magic angle spinning (MAS) NMR spectroscopy of as-made
samples were carried out at room temperature on a Bruker AV-400-
WB equipment (details are given in the Supporting Information).
The amount of organic matter (SDA) occluded in the zeolite was
determined by C, H, N elemental chemical microanalysis in a LECO-
CNNS-932 analyzer. Thermogravimetric analyses were performed
under oxygen flow (100 mL/min) in an SDT Q600 TA Instruments
equipment up to 1000 °C (with a heating rate of 10 °C/min).
Molecular Mechanics Simulations. The location and interaction
energies of the different imidazolium derivatives in MFI and TON
were studied by molecular mechanics simulations, as implemented in
the Forcite module in the Materials Studio software.23 The geometry
of the zeolite structures has been kept fixed during all these
calculations. Molecular structures and the interaction energies of the
organic SDAs with the framework were described with the CVFF
forcefield.24 Periodic boundary conditions (PBC) were applied in all
the calculations. The atomic charges for the organic molecules were
calculated by the charge-equilibration method,25 setting the total net
molecular charge to +1 (1E3MI) or +2 (for 3BI, 4BI, 5BI, and 6BI).
The positive charge of the organic SDA molecules was compensated
by the framework using a related version of the uniform charge back-
ground method,26 where the atomic charge for every silicon frame-
work atom was reduced from 0.6 until charge neutrality. Framework
oxygen charges were kept fixed to −0.3. These framework charges
were used in order to simulate the hydrophobicity of the all-silica
ZSM-5 material. We also studied the effect of introducing localized −1
charges on F atoms, with F located in the [415262] cages of the MFI
structure, in the same positions as revealed by XRD in an MFI sample
obtained with tetrapropylammonium and fluoride;27 in this mixed
charge-background model, half of the cationic charges were
compensated by the presence of F− anions (4), and the other half
by a uniform charge background on Si atoms (corresponding to 4
SiOH/SiO− defects, as experimentally observed, see below). We found
exactly the same trend in the interaction energies as with the initial
uniform charge-background method (see comparison below), showing
that the latter is a good approximation for charge-balancing purposes.
Initially, the molecules were manually docked in the structures in
appropriate supercells of the zeolite frameworks under packing values
closely similar to those observed experimentally. The most stable loca-
tion for the SDA molecules was obtained by means of simulated an-
nealing calculations. The interaction energy was calculated by subtracting
the energy of the molecules in vacuo to the total energy of the system,
and was normalized to energies per unit cell and per Si atom; all the
energy values are given in kilocalories per mole (kcal/mol).
EXPERIMENTAL SECTION
■
Synthesis of the SDAs. The organic SDAs used in this work
comprise a series of doubly charged cations consisting of two N-
methylimidazolium charged moieties spaced by a linear bridge of n = 3,
4, 5, or 6 methylene groups, plus the singly charged N-ethyl-N′-
methylimidazolium (Chart 1). The doubly charged SDAs, denoted as
Chart 1. Organic Cations Used in This Work as SDAs
nBI, were synthesized by reaction of 1-methylimidazole with the
corresponding linear α,ω-dibromoalkane, Br(CH2)nBr. In a typical
synthesis, 11.494 g (0.14 mol) of 1-methylimidazole (Aldrich 98%)
was dissolved in 100 mL of chloroform (Aldrich 99%). The linear alkyl
bromide (0.07 mol, Fluka 99% for n = 3 or Aldrich 96−99% for n = 4−6)
was added to the solution, which was kept under magnetic stirring at
room temperature for 4 days. The bromide salt was then separated by
rotoevaporation. The singly charged 1-ethyl-3-methylimidazolium,
denoted 1E3MI, was bought as its chloride salt from Aldrich (>95%).
The 1H NMR spectra of the SDA salts dissolved in D2O or CDCl3
confirmed that we obtained the different SDAs with acceptable purity
(Supporting Information Figure S1). When the spectra were recorded
in D2O, the moderately acidic H at position 2 (labeled as ‘a’ in Figure S1)
of the imidazolium ring eventually showed a reduced relative intensity due
1
to H−2H exchange.
The dibromide or chloride salts were converted into the corre-
sponding hydroxides by anion exchange. A total of 0.03 mol of SDA
salt was dissolved in water (1 SDA/10 H2O weight ratio) and the
solution was contacted with 100 mL of exchange resin (Dowex mono-
sphere 550A (OH) anion exchange resin, Aldrich, 1.1 mequiv./1 mL).
After 12 h under stirring at room temperature, the hydroxide solution
was recovered by filtration, and was concentrated by rotoevaporation
under vacuum at 414 K. The final concentration of the hydroxide solu-
tion was determined by titration with HCl 0.1 N. The yield of the
exchange plus concentration process (final equivalents of hydroxide
per 100 starting equivalents of cation) was over 90%.
Density Functional Theory Calculations. To understand the 19F
NMR spectra of the MFI samples observed experimentally, a com-
plementary computational study based on quantum mechanics was
performed. Calculations were carried out within the Density Functio-
nal Theory (DFT) methodology, using plane-waves as basis set (with a
cutoff of 500 eV), and the PBE generalized gradient approximation as
functional.28 Calculation of the NMR parameters were performed with
the gauge-including projector augmented-wave method (GIPAW)
developed by Pickard and Mauri,29 as implemented in the CASTEP30
module in Materials Studio.31 This computational methodology has
been shown to accurately predict the NMR properties of zeolite
materials.32 Dispersion interactions were accounted for through the
Grimme dispersion method (DFT+D).33 The MFI structure with the
SDA molecules and F ions was first geometry optimized under this
level of theory, and subsequently, the NMR properties were calculated,
Synthesis of Zeolites. Tetraethylorthosilicate (TEOS, 98%,
Aldrich) was hydrolyzed under stirring at room temperature in an
aqueous solution of the hydroxide form of the corresponding SDA
cation. Some water and all ethanol, produced by the hydrolysis, were
eliminated by evaporation, which was monitored by weight loss. The
stirring was stopped when the desired H2O/SiO2 ratio was achieved.
Then, the required amount of a concentrated aqueous solution of HF
(48% Aldrich) was added while stirring with a spatula for 15 min.
The obtained gels were distributed in Teflon lined stainless steel
autoclaves which were heated at 423 K while tumbled at 60 rpm.
3846
dx.doi.org/10.1021/ja210703y | J. Am. Chem. Soc. 2012, 134, 3845−3856

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Journal of the American Chemical Society
Article
obtaining the absolute shieldings of the different nuclei. The chemical
shift for a nucleus in a given position (δ(r)) is then defined as:
thus, may be crystallized using many different organics in a
wide range of conditions. Alternatively, one can say that when
an SDA fails to display any significant specificity in its structure-
direction role, a default structure able to loosely accommodate
the organics crystallizes.
δ(r) = σ − σ(r)
(1)
ref
where σ(r) is the isotropic shielding obtained in the calculations. To
compare with the experimental 19F chemical shifts, we selected a σref of
137 ppm, in such a way that the theoretical and experimental chemical
shifts of the most shielded F atom coincide at ∼−80 ppm. Indeed, we
are successfully using the same value of σref in other ongoing studies of
19F NMR properties of other fluorine-containing microporous solids.
In this work, the only hint for a more specific direction
among the bis-imidazolium series is displayed by the 4BI cation
that, in addition to TON, was also able to produce zeolite MFI
when the crystallization was carried out at very high concen-
tration (H2O/SiO2 ratio of 4.8). The fact that a four-methylene
spacer produces zeolite MFI is in sharp contrast with previous
reports showing that hexamethylene bridges in polyalkylam-
monium SDAs are generally more adequate to yield this
zeolite.22,34 For polycations with cationic moieties located at
channel intersections of MFI and separated by linear alkane
spacers, it has been found that 6 methylene spacers apparently
closely match the distance between adjacent channel
intersections, allowing an optimal fit, while 5 or 7 methylenes
may also eventually work.22,35 This suggests that the two
imidazolium rings of each 4BI molecule do not occupy adjacent
channel intersections. Interestingly, the singly charged 1E3MI,
which can be formally considered as the “monomer” of the 4BI
bis-imidazolium, showed a similar structure-direction effect,
yielding generally TON, and only when working at very high
concentrations (H2O/SiO2 ratio of 3.5) also MFI. Even in
those cases, MFI was observed to be metastable and trans-
formed into the more stable TON by Ostwald ripening. It is
also worth noting that the crystallization of MFI, and its trans-
formation to TON, is much faster with 4BI than with 1E3MI
(Table 1).
The observation that 4BI and 1E3MI may yield MFI or TON
depending on the degree of concentration of the synthesis
mixture agrees with the general observation, recently termed as
the Villaescusa’s rule,36 that silica materials with a lower frame-
work density are favored at more concentrated conditions in
fluoride media.37 Additionally, since the rest of cations used in
this work gave no hint of MFI even at high concentration
and short crystallization time, apparently a faint but certain
structure-direction effect as the one sought exists. This merited,
in our opinion, a theoretical investigation, which in turn re-
quired a detailed characterization of the obtained materials,
specially of the ones crystallized with 4BI and 1E3MI.
RESULTS AND DISCUSSION
Zeolite Synthesis. All the SDAs used in this work con-
sistently produced TON (Table 1), a zeolite that can be
■
Table 1. Summary of Synthesis Results
SDA
H2O/SiO2
8.5
time (days)
phase
1E3MI
3
6
TON
TON
TON
18
5
3.5
Amorphous
MFI
7
17
19
3
TON + MFI
TON
3BI
7.1
4.3
6.9
Amorphous
TON + amorphous
Dense phase
Amorphous
TON
6
13
1
3
9
2
TON
TON
4BI
5BI
6BI
5
TON
26
1
TON
4.8
6.4
4.3
MFI
5
4
TON
TON
8
TON
1
Amorphous
Amorphous
TON
TON + amorphous
TON
8
15
3
6.0
4.3
6
12
2
TON
Zeolite Characterization. Table 2 lists chemical compo-
sition details of each kind of SDA-zeolite obtained in this work.
The C/N molar ratio of the zeolites is close to the value in the
SDA used suggesting that the SDAs are occluded chiefly intact
within the zeolites, although there are deviations that suggest
decomposition products might also get occluded in certain
Amorphous+TON
TON
13
considered as a default structure for pure silica compositions.
A default structure12 may be defined as one that does not require
any particularly strong specificity in structure direction and,
Table 2. Chemical Composition of the Zeolites Obtained in This Work
a
a
b
SDA
phase
%C
%H
%N
C/N
H/N
empirical formula
1E3MI
4BI
MFI
5.1
7.5
4.2
4.6
5.4
4.5
4.6
0.9
1.3
0.8
0.8
1.0
0.9
0.9
1.9
3.2
1.7
2.2
2.2
1.9
1.7
3.1(3.0)
2.7(3.0)
2.9(3.0)
2.4(2.75)
2.9(3.0)
2.7(3.25)
3.1(3.5)
6.6(5.5)
5.6(5.0)
6.5(5.5)
5.1(4.5)
6.3(5.0)
6.6(5.5)
7.4(6.0)
|C6H11N2F|4.4[SiO2]96:5.30H2O
|C12H20N4F2|3.9[SiO2]96:4.50 H2O
|C6H11N2Fx|0.96[SiO2]24:1.00 H2O
|C11H18N4Fx|0.62[SiO2]24:0.70 H2O
|C12H20N4Fx|0.64[SiO2]24:1.69 H2O
|C13H22N4Fx|0.54[SiO2]24:1.18 H2O
|C14H24N4Fx|0.49[SiO2]24:1.28 H2O
MFI
1E3MI
3BI
TON
TON
TON
TON
TON
4BI
5BI
6BI
a
b
Molar ratio. The theoretical value is given in parentheses. SDA from N analysis assuming the SDA is intact, SiO2 taken as the TG residue and
water derived from the excess H/N ratio (it may actually be water or SiOH defects). The amount of F in 4BI-MFI is derived from the relative
intensity of the 29Si resonances around −145 ppm, assigned to [SiO4/2F]− units (see text). For the TON zeolites, a significant concentration of
connectivity defects is revealed by 29Si MAS NMR and the amount of F balancing cationic charges is not determined.
3847
dx.doi.org/10.1021/ja210703y | J. Am. Chem. Soc. 2012, 134, 3845−3856

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Journal of the American Chemical Society
Article
cases. This is more conclusively deduced from the 13C MAS
NMR spectra of the materials, which show the expected re-
sonances for the alkyl (50−0 ppm) and aromatic (140−
120 ppm) carbons, although, at least in the 3BI-TON sample,
additional resonances likely due to decomposition products are
also observed (Supporting Information Figure S2). Apparently,
trimethylene bridges may provide a reduced stability of doubly
charged SDAs, since Lee et al. also reported degradation of tri-
methylene-bis(trimethylammonium) in conditions comfortably
withstood by larger diquats.14
interaction in the materials prepared with 1E3MI and 4BI.
While the most prominent features in the spectra of both
materials are the Si Q4 resonances in the −108/−120 ppm
region, a very broad band centered around −125 ppm is observed
in 1E3MI-MFI, which is characteristic of Si sites changing their
coordination between 4 and 5 ([SiO4/2] and [SiO4/2F]−,
respectively), a situation similar to that found in TPA-MFI
synthesized in fluoride medium.38 In contrast, this resonance is
absent in the spectrum of 4BI-MFI, which shows instead two
sharp and small resonances around −145 ppm, whose relative
intensity is sharply increased in the 29Si{19F} CP MAS NMR
spectrum. This is indicative of pentacoordinated [SiO4/2F]− units
which are observed in TPA-MFI only when the dynamic situation
described above is frozen at 140 K.39 A deconvolution of the 29Si
MAS NMR spectrum of 4BI-MFI indicated that around 3.3% of
the Si atoms (3.2 Si per unit cell of 96 Si) are involved in
[SiO4/2F]− units. Since, according to CHN analysis, there are
almost four doubly charged 4BI cations per unit cell, this suggests
that barely half of the cationic charges are compensated by
[SiO4/2F]− units. The 29Si MAS NMR spectrum also shows that,
contrarily to 1E3MI-MFI and the most general case for pure silica
zeolites synthesized in fluoride medium, which typically have a
very small concentration of defects,40 the sample crystallized with
4BI contains a significant amount of Q3 defects evidenced by the
broad resonance around −103 ppm (13.6%, or 13.1 SiO− or
SiOH per unit cell). This suggests that balance of the SDA
cationic charges in 4BI-MFI involves both fluoride and SiO− and
that the total amount of Q3 largely exceeds (around 4 times) that
of SiO−. This Q3 excess is similar to that typically found in silica
zeolites prepared in highly basic conditions using cationic SDAs.41
As-made TON materials show a strong resonance around
−112 ppm due to Q4 sites together with a weaker one around
−102 ppm characteristic of Q3 (Si−O− or Si-OH defects,
spectra not shown). There is no clear indication in these
spectra of silicon sites interacting with fluoride, although 19F
MAS NMR show resonances in the −75/−76 ppm typical
range for SiO4/2F− in TON.7
The 13C NMR spectra of 4BI and 1E3MI in MFI or TON
zeolites (Figure 1) show a close similarity, as expected due to
Figure 1. 13C{1H} CPMAS NMR spectra of as-made TON and MFI
zeolites synthesized with 1E3MI (bottom) and 4BI (top).
their similar molecular structure. For both cations, the aromatic
carbons at positions 4 and 5 of the imidazolium rings are better
resolved in TON than in MFI zeolites, which may point to a
wider range of different environments for the imidazolium ring
in MFI resulting in broader resonances. Interestingly, the orga-
nic contents are much larger in the MFI sample synthesized
with 4BI than in the one prepared with 1E3MI (Table 2).
The 29Si MAS NMR and 29Si{19F} CP MAS NMR spectra of
the MFI as-made materials (Figure 2) evidence a different Si−F
Figure 3 shows the 19F MAS NMR spectra of the as-made
MFI zeolites. Both samples contain two resonances: one
Figure 3. 19F MAS NMR spectra of the as-made 1E3MI-MFI (top)
and 4BI-MFI (bottom).
around −66 or −68 ppm (for 1E3MI-MFI and 4BI-MFI,
respectively), appears at a chemical shift similar to that
observed in TPA-F-MFI (−64 ppm),39 and the other one is
at a higher field (−79 ppm in both cases). In 4BI-MFI, both
resonances are broad and have almost exactly the same
Figure 2. 29Si MAS NMR and 29Si{19F} CPMAS NMR spectra of the
as-made MFI materials.
3848
dx.doi.org/10.1021/ja210703y | J. Am. Chem. Soc. 2012, 134, 3845−3856