Multitechnique experimental insight on an unusual crystal-to-crystal high temperature solid state reaction in zirconium carboxypyridinephosphonates: From one-dimensional chains to two-dimensional hybrid layers through HF elimination

Article abstract of DOI:10.1021/cg301042y

Two novel N-phosphonoethylcarboxypyridines were prepared via nucleophilic substitution of two carboxypyridines (namely, 4-carboxypyridine, or isonicotinic acid, and 3-carboxypyridine, or nicotinic acid) with diethyl-2- bromoethylphosphonate in water. Two zirconium derivatives of these acids were obtained under mild solvothermal conditions, and their structures were solved from powder X-ray diffraction (PXRD) data. The zirconium derivative containing the isonicotinic moiety (1), with formula ZrF₂(HF)(O ₃PCH₂CH₂NC₅H₄CO ₂), has a one-dimensional (1D) chain structure in which the carboxypyridine groups are placed in the external part of the inorganic chain constituted of zirconium octahedra and phosphonic tetrahedra, whereas the zirconium derivative containing the nicotinic moiety (2), with formula ZrF ₂(O₃PCH₂CH₂NC₅H ₄CO₂), has a hybrid layered structure in which zirconium octhaedra and phosphorus tetrahedra form a new structural archetype, with a C-O group coordinating the zirconium atoms. 1 underwent a high temperature (280 °C) slow solid state transformation that involved the loss of one HF molecule coordinated to the zirconium atom and the replacement of this coordination vacancy with the neighboring C-O^- group belonging to the adjacent chain. The structure of this heated compound (1a) is a polymorph of 2. A multitechnique approach, based on coupled Fourier transform infrared spectroscopy and Raman and solid state NMR spectroscopy allowed us to carry out a thorough characterization of these materials, finding nice agreements on the chemical details of this solid state reaction.

Full text of DOI:10.1021/cg301042y

Article
pubs.acs.org/crystal
Multitechnique Experimental Insight on an Unusual Crystal-to-
Crystal High Temperature Solid State Reaction in Zirconium
Carboxypyridinephosphonates: From One-Dimensional Chains to
Two-Dimensional Hybrid Layers Through HF Elimination
Ferdinando Costantino,†,⊥ Paola Sassi,† Marco Geppi,‡ and Marco Taddei*,†
†Department of Chemistry, University of Perugia, Via Elce di sotto n. 8. 06123 Perugia, Italy
‡Department of Chemistry and Industrial Chemistry, University of Pisa, Via Risorgimento 35 - 56126 Pisa, Italy
⊥ICCOM-CNR, Area di Ricerca CNR, Firenze, Via Madonna del Piano 10, 50019 Sesto Fiorentino, Italy
S
* Supporting Information
ABSTRACT: Two novel N-phosphonoethylcarboxypyridines
were prepared via nucleophilic substitution of two carbox-
ypyridines (namely, 4-carboxypyridine, or isonicotinic acid,
and 3-carboxypyridine, or nicotinic acid) with diethyl-2-
bromoethylphosphonate in water. Two zirconium derivatives
of these acids were obtained under mild solvothermal
conditions, and their structures were solved from powder X-
ray diffraction (PXRD) data. The zirconium derivative
containing the isonicotinic moiety (1), with formula ZrF2(HF)(O3PCH2CH2NC5H4CO2), has a one-dimensional (1D) chain
structure in which the carboxypyridine groups are placed in the external part of the inorganic chain constituted of zirconium
octahedra and phosphonic tetrahedra, whereas the zirconium derivative containing the nicotinic moiety (2), with formula
ZrF2(O3PCH2CH2NC5H4CO2), has a hybrid layered structure in which zirconium octhaedra and phosphorus tetrahedra form a
new structural archetype, with a C−O group coordinating the zirconium atoms. 1 underwent a high temperature (280 °C) slow
solid state transformation that involved the loss of one HF molecule coordinated to the zirconium atom and the replacement of
this coordination vacancy with the neighboring C−O− group belonging to the adjacent chain. The structure of this heated
compound (1a) is a polymorph of 2. A multitechnique approach, based on coupled Fourier transform infrared spectroscopy and
Raman and solid state NMR spectroscopy allowed us to carry out a thorough characterization of these materials, finding nice
agreements on the chemical details of this solid state reaction.
architectures with very different dimensionality and connectiv-
ity.
INTRODUCTION
■
The chemistry of zirconium phosphonates has its roots in the
1970s, but still today, after more than 30 years, it continues to
give rise to a remarkable interest in the scientific community,
not only from a fundamental point of view, but also for the
extent of applications in which they are still employed.1
A peculiar feature of these materials resides in their
considerable chemical and thermal stability accompanied with
a high structural variability. The latter is mainly due to the ease
of modulating their structure and reactivity through the rational
choice of the phosphonic building blocks. Such double
characteristics have been recognized, both in the past and still
today, as a fundamental aspect for their application in several
fields like proton conductivity, heterogeneous catalysis, fillers
for polymeric composites, thin films, and so on.2
The ways to assemble zirconium atoms and polyphosphonic
groups are almost countless because the capability of
phosphonate groups of acting as mono- or divalent and
mono-, di-, or tridentate ligands, and the presence of
noncovalent interactions, such as H-bonds and hydrophobic/-
philic interactions, play a crucial role in building up structural
In the last years an important contribution to this chemistry
was given by our research group with the publication of several
zirconium derivatives with one-, two-, and three-dimensional
(1D, 2D, and 3D) structures based on aminomethylenephos-
phonic acids obtained from the reaction of aliphatic amines,
phosphorous acid, and formaldehyde.3
Alternatively, phosphonic groups can be covalently linked to
aromatic heterocycles (such as pyridines, pyrazoles, and similar
molecules) by using a method based on the nucleophilic
substitution of these ones on diethyl 2-bromoethylphosphonate
in aqueous medium and subsequent acid hydrolysis. This
reaction produces a charged quaternary nitrogen atom balanced
by a counterion (normally Cl− or Br−, depending on the acid
used for hydrolysis) which could be incorporated into the
structure. Pioneering studies in similar systems have been
Received: July 23, 2012
Revised: September 14, 2012
© XXXX American Chemical Society
A
dx.doi.org/10.1021/cg301042y | Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
carried out by Vermeulen and co-workers in the beginning of
1990s, when a series of zirconium phosphonates containing
pyridine and viologen groups have been prepared and their
photophysical and photocatalytic properties studied in fine
detail.4
In the present paper we report the synthesis of two novel
phosphono-carboxylate ligands carried out by the reaction of
diethyl 2-bromoethylphosphonic with 3-carboxypyridine (nic-
otinic acid) and 4-carboxypyridine (isonicotinic acid). The
molecular formula of these two novel ligands is
[H2O3PCH2CH2NC5H4CO2H]+Cl− (H3L1 for isonicotinic
and H3L2 for nicotinic) and their structures are depicted in
Chart 1.
EXPERIMENTAL SECTION
■
Reagents. ZrOCl2·8H2O was a Merck Pro Analysi product. All the
other chemicals were purchased from Sigma-Aldrich.
Synthesis of the Phosphonic Acids. The N-phosphonoethyl-4-
carboxypyridinium chloride (H3L1) was synthesized by modifying the
procedure reported in the literature:4a,b One gram of isonicotinic acid
(8.1 mmol) was mixed with 15 mL of NH3 0.54 M (8.1 mmol), and 3
g of diethyl-2-bromoethylphosphonate (12.1 mmol) was added to this
solution. The mixture was heated at 90 °C for 6 days, and then 15 mL
of 12 M HCl was added and the mixture was left reacting for another
day. At the end of the reaction the solvent was evaporated under a
vacuum and the residue was treated with isopropanol. A white solid
precipitated and was filtered. This solid was a mixture of NH4Cl and
H3L1, for a total weight of 1.7 g. Titration of the solid with NaOH
revealed that the amount of H3L1 in this mixture was 70% (1.191 g).
Yield: 55%.
Chart 1. Molecular Formulas of Chloride Salts of H3L1 and
H3L2
1H NMR (D2O, 400 MHz): δ = 8.96 (d, 2H, aromatic, 2,6-
position), 8.34 (d, 2H, aromatic, 3,5-position), 4.78 (m, 2H N−CH2-),
2.32 (m, 2H, −CH2−P) ppm. 31P NMR (D2O, 400 MHz): δ = 19.32
(m) ppm.
The N-phosphonoethyl-3-carboxypyridinium chloride (H3L2) was
prepared with a similar procedure, except for the workup of the
product: after evaporation of the solvent, the residue was treated with
a 9:1 isopropanol/anhydrous ethanol mixture. A white solid
precipitated and was filtered. This solid was a mixture of NH4Cl, 3-
carboxypirydinium chloride, and H3L2, for a total weight of 1.5 g.
Titration of the solid with NaOH revealed that the amount of H3L2 in
this mixture was 66% (0.990 g). Yield: 46%.
1H NMR (D2O, 400 MHz): δ = 9.4 (s, 1H, aromatic, 2-position)
9.03 (d, 1H, aromatic, 6-position), 8.94 (d, 1H, aromatic, 4-position),
8.13 (m, 1H, aromatic, 5-position) 4.85 (m, 2H N−CH2-), 2.37 (m,
2H, −CH2−P) ppm. 31P NMR (D2O, 400 MHz): δ = 18.92 (m) ppm.
Synthesis of the Zirconium Derivatives. Zr derivatives of H3L1
and H3L2 of formula ZrF2(HF)(PO3CH2CH2NC5H4CO2) (hereafter
1) and ZrF2(PO3CH2CH2NC5H4CO2) (hereafter 2), respectively,
were prepared as follows: 5 mL of a clear solution of the phosphonic
acid (0.5 mmol) were mixed with a solution obtained by mixing
ZrOCl2·8H2O (0.5 mmol) and 0.7 mL of HF 2.9 M (2 mmol). These
solutions were maintained in closed plastic vessels for 3 days at 80 °C.
The solids obtained were filtered under a vacuum, washed with water,
and dried in an oven at 70 °C. Yields (calculated on Zr): 71% for 1,
53% for 2.
Analysis: Calcd for ZrPC8F3NO5H10 (1) Zr = 23.95%, P = 8.15%, F
= 14.99% . Exp: Zr = 23.43%, P = 7.88%, F = 16.6%. Calcd for
ZrPC8F2NO5H9 (2) Zr = 25.33, P = 8.86%, F = 10.58%. Exp: Zr =
24.09, P = 8.68%, F = 9.75%.
Thermal Synthesis of 1a. Compound 1a was obtained by heating
compound 1 at 280 °C for 24 h in a platinum crucible. During this
time the structural transformations were followed by PXRD. The
thermal treatment was concluded only when the peaks belonging to
the pristine phase completely disappeared.
Analytical and Instrumental Procedures. The zirconium and
phosphorus contents of the samples were obtained by inductively
coupled plasma optical emission spectroscopy using a Varian Liberty
Series II instrument working in axial geometry, after mineralization of
the samples with hydrofluoric acid.
The fluorine content was determined as fluoride by ion
chromatography: about 0.1 g of sample was refluxed for 3 h with 10
mL of 1 M NaOH up to complete hydrolysis. The resulting solution
was filtered, properly diluted, and analyzed with a Dionex series 2000
i/sp instrument, using an IonPack AS4A column and a buffer solution,
of the following composition: 1.7 × 10−3 M in NaHCO3 and 3.5 ×
10−3 M in Na2CO3 as eluent.
Two zirconium derivatives of [H3L1] and [H3L2] were
synthesized under mild solvothermal conditions, and their
structures were solved ab initio from powder X-ray diffraction
data (PXRD). By reacting [H3L1] with zirconium in hydro-
fluoric acid media, a compound with formula ZrF2(HF)(L1)
(hereafter 1) was obtained, which shows a 1D chain structure
constituted of an inorganic backbone of ZrO3F3 octahedra and
PO3C tetrahedra and with the pending aromatic isonicotinic
groups hydrogen-bonded to the adjacent chains. The reaction
of [H3L2] with zirconium yielded a compound with formula
ZrF2(L2) (hereafter 2), having a layered hybrid structure in
which the layers are constituted by 1D chains of ZrO4F2
octahedra and PO3C tetrahedra with a connectivity never
observed before, linked to each other through the carboxylate
groups of the nicotinic moieties.
When heated at 280 °C, 1 underwent a very slow structural
transformation that was triggered by the loss of one of the
fluorine ligands linked to zirconium as HF, with the consequent
formation of a coordination vacancy on the zirconium sphere
and the subsequent condensation of two adjacent chains
through the formation of a new Zr−O bond filling the
coordination vacancy. This new compound (1a) is a perfect
polymorph of 2, despite the important differences of the
inorganic chain connectivity.
Useful information about the structural transformation
observed by PXRD was provided by a thorough investigation
by means of solid-state Fourier transform infrared (FT-IR)
spectroscopy, Raman, and NMR spectroscopy, especially
focusing on the local environment of the zirconium atom.
The data collected by means of these techniques were in
excellent agreement with the observations from PXRD, thus
allowing us to have a detailed picture of the structural features
of the investigated compounds.
PXRD patterns for structure determinations and Rietveld refine-
ments were collected with a PANalytical X’PERT PRO diffractometer
equipped with a X’Celerator solid state fast detector in the 3−100° 2θ
range and with a 30 s/step counting time.
B
dx.doi.org/10.1021/cg301042y | Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
Thermogravimetric analyses (TGA) were performed using a
Netzsch STA490C thermoanalyzer under a 20 mL min−1 air flux
with a heating rate of 5 °C min−1.
Table 1. Structural Parameters and Refinement Details for 1,
1a, and 2
compound
empirical
1
1a
2
Raman spectra were excited using red radiation from a He−Ne
laser; the laser power on the sample was always less than 5 mW.
Backscattering illumination and collection of the scattered light were
made through an Olympus confocal microscope MOD BX40 (50x and
100× objectives). When a 1800 lines mm−1 grating was used, a spectral
resolution of about 2 cm−1 was achieved. Spectral measurements were
made with continuous scans in the range 200−1800 cm−1, exposure
times in the range 30−60 s, and 10 accumulations. Detection of
scattered light was accomplished by a 1024 × 128 CCD array, mod.
SPECTRUM-ONE by Horiba - Jobin Yvon. Calibration of the
spectrometer was accomplished using the Raman lines of a polystyrene
standard.
Mid-FT-IR measurements were carried out with a Bruker
TENSOR27 spectrophotometer composed of a globar source, a high
throughput patented RockSolid cube corner interferometer, a
deuterated L-alanine tryglycine sulfate detector and a KBr beamsplitter.
Transmission spectra were collected at room temperature in the 400−
4000 cm−1 range. Each spectrum was the average of 50 scans,
measured with a resolution of 2 cm−1. The samples were dispersed
into anhydrous KBr pellets.
ZrPC8F3NO5H10
ZrPC8F2NO5H9
ZrPC8F2NO5H9
formula
formula weight
crystal system
space group
a (Å)
378.28
358.28
358.28
monoclinic
P21/a
monoclinic
P21/a
monoclinic
P21/a
14.2649(3)
12.8066(2)
6.6816(1)
102.447(3)
1191.93(4)
4
12.8478(9)
12.8510(5)
6.6366(6)
90.386(9)
1095.7(1)
4
17.6407(8)
13.0625(6)
9.9599(5)
97.727(6)
2274.2(2)
8
b (Å)
c (Å)
β (deg)
volume (Å3)
Z
T (°C)
25
25
25
calc density/
2.06
2.12
2.04
cm−3
no. of data
6764
3643
6577
3802
6640
3386
no. of
reflections
no. of variables
80
62
132
Solid state NMR spectra were recorded on a two-channel Varian
Infinity Plus 400 spectrometer working at 400.03 and 376.37 MHz for
1H and 19F nuclei, respectively, equipped with a cross-polarization
magic angle spinning (CP-MAS) probe bearing 3.2 mm (outer
no. of restraints 40
65
132
Rp
0.054
0.068
0.094
0.145
4.84
0.046
0.063
0.109
8.85
Rwp
0.076
0.103
3.33
2
RF
1
diameter) rotors. H and 19F 90° pulse durations were 2.1 and 3.6 μs,
χ
respectively. Optimized relaxation delays of 5 and 30 s were used, and
1
a number of scans of 16 and 40 were accumulated for H and 19F,
respectively. 1H spectra were recorded using a single-pulse experiment,
while the Hahn Echo pulse sequence was used for 19F MAS
experiments with an echo delay of 15 μs. Spectra at different MAS
frequencies were recorded between 5 and 23 kHz. The chemical shift
scales of 19F and 1H spectra are referenced to the signals of CFCl3 and
TMS, respectively. All the spectra were recorded at 21 °C.
The statistical weight of these restraints was decreased as the
refinement proceeded. For 1 the atomic displacement parameters
were refined constraining the program to apply the same shift except
Zr atoms, whereas those of 1a and 2 were not refined. At the end of
the refinement, the shifts in all parameters were less than their
standard deviations.
Structure Determination and Refinement for 1, 1a, and 2. A
preliminary profile fit was conducted on the diffraction patterns of 1,
1a, and 2 using a split Pearson VII profile function. The position of the
first 20 lines (Kα1 maxima) was determined and used for indexing
with the TREOR905 program. Space groups were assigned on the basis
of a systematic comparison of the number of peaks found and the
number of possible peaks, within all space groups of the respective
crystal symmetry, using the Chekcell6 program.
Notably, the cell volume of 2 is twice than that of 1 despite the
similarities of the two compounds in terms of molecular weight. As a
matter of fact, the asymmetric unit of 2 found after the structure
determination process is also doubled, and there are eight formula
units per unit cell. Several attempts to find a reduced cell failed and the
structure was determined by using the larger unit cell.
The structures of the three compounds were solved using global
optimization methods (FOX7 program). The FOX program is able to
optimize in direct space the molecules described in terms of bond
lengths and angles. Optimization is performed over a cost function, in
our case the integrated Rwp. The reverse Montecarlo parallel tempering
algorithm8 was used, and antibump distances between metals and
other atoms were also added. A ZrO6 octahedron with distances Zr−O
of 2.05 Å with standard deviations of 0.05 Å was used. The molecules
of H3L1 and H3L2 ligands were described in terms of bond lengths and
angles set at their literature values with standard deviations of 0.15 Å
and 10°, respectively. For the structural determination of 1a several
trials in tetragonal symmetry have been performed, but no reliable
structural model has been found. The real structure has been found
using the pseudotetragonal cell with monoclinic symmetry whose cell
parameters are shown in Table 1.
RESULTS AND DISCUSSION
■
Crystallography. Crystal data and details of the refinement
for 1, 1a, and 2 are reported in Table 1. Figure 1 shows the final
Rietveld and difference plots.
The structure of 1 is depicted in Figure 2. It has a 1D chain
structure built from an inorganic macro-unit constituted of
PO3C tetrahedra and ZrO3F3 octahedra running along the c-
axis. This unit is constructed by the repetition of a four-
membered ring of P−O−Zr−O−P atoms that can be also
intended as a composite building unit (hereafter CBU) of two
Zr octahedra and two P tetrahedra. The same connectivity was
observed in other zirconium phosphonates, namely, the
zirconium N,N′-bis(phosphonoethyl)-4,4′-bipyridinium (Zr−
P2Viologen)4b and more recently in two zirconium amino-
phosphonates containing piperidine and bis-piperidine
groups.3f Each phosphonic tetrahedron is triply connected to
three different zirconium atoms, and each zirconium is thus
connected to three oxygen atoms belonging to three different
phosphonic groups and to three terminal fluorine atoms. Two
fluorine atoms are placed in axial position, with the third one
occupying the equatorial plane in a mer (or trans) reciprocal
configuration.
The isonicotinic groups are placed in the external part of the
chain through the linking phosphonoethyl groups and, because
of the disposition of the phosphonic tetrahedra, they are also
alternatively placed up and down with respect to the plane
defined by the zirconium atoms. Each COO group of the
isonicotinate points toward an adjacent chain and one of the
two carboxylate oxygens is near an equatorial fluorine of the
Structural refinement of 1, 1a, and 2 was performed using the
Rietveld method implemented in the GSAS9 program. Cell parameters,
sample displacements, background, and profile were first refined, and
then atomic coordinates and atomic displacement parameters were
also refined. Soft restraints were used for bond distances and angles.
C
dx.doi.org/10.1021/cg301042y | Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
Figure 2. Polyhedral representation of the structure of 1 viewed down
the c-axis (a), and a detail of a single layer (b). ZrO3F3 octahedra are
represented in purple, and PO3C tetrahedra are represented in green.
Hydrogen bonds are represented as red dashed lines.
Tetrahedra and octahedra are assembled as follows: there are
the same composite building units of compound 1 (represented
in light blue in Figure 5), defined by a four-membered ring,
which are staggered up and down in a corrugated fashion
(Figure 4a,b). The main difference with respect to 1 (Figure
4c) is that in this compound the CBUs are connected through
the equatorial oxygen atoms of the zirconium sphere, and the
two zirconium atoms of each CBU are crystallographically and
chemically equivalent. For 2, the CBUs are connected through
the apical oxygen atoms of Zr octahedra. Moreover, these are
not equivalent for symmetry and the asymmetric unit is twice
that of 1. This difference can be ascribed to the cis- disposition
of the F atoms, one of which (F2 in the case of Zr1 and F4 in
the case of Zr2) occupies in 2 the same position occupied by an
oxygen atom of the P-tetrahedra and therefore breaks the
propagation of the CBU along the c-axis, forcing a corrugated
disposition.
Like in 1, also in 2 the nicotinic groups are placed in the
external part of the chains, but, at variance with 1, the chains are
covalently linked to each other by means of the bond between
the carboxylate group and the zirconium atom. In fact, such a
bond substitutes the noncovalent C−O···F interaction found in
compound 1. As a consequence, in this compound the
counterion balancing the positive nitrogen is represented by
the ZrO4F2 complex, bearing a formal negative charge. Notably,
no examples of zirconium carboxyphosphonates obtained by
direct synthesis with a carboxylic group coordinating the
zirconium atom existed up to now: the only example reported
with such a feature was obtained by postsynthetic reaction.3b
The above-described structural features bring a global
structure in which the layers have a considerable thickness
Figure 1. Rietveld plots for 1, 1a, and 2.
neighboring Zr octahedra with a distance of C8−O4···F2 of
2.67(2) Å. This short distance strongly suggests that a
noncovalent interaction is present between these two atoms
or, more precisely, that F is protonated by the COOH group.
PXRD data do not provide any information about the position
of the hydrogen atom involved in this interaction, but useful
information can be derived from the IR, Raman, and NMR data
that will be discussed later. Adjacent chains are linked to each
other through these noncovalent interactions and lie along the
a-axis. No guest species have been found from different Fourier
calculations suggesting that no place is available among the
chains. In particular, there are no extraframework counterions
balancing the positive charge on the nitrogen atoms, since this
role is played by the deprotonated carboxylic groups.
The structure of 2 is depicted in Figure 3. It has a layered
structure and the layers are constituted by the assembly of
inorganic macro-units linked to each other, along the layer
plane, by the nicotinic groups. This linkage is allowed by the
coordination on zirconium atoms of oxygen atoms belonging to
the carboxylic groups so that zirconium octahedra have three
oxygen atoms coming from the phosphonic tetrahedra, one
from the carboxylic group and two terminal fluorine atoms in
cis position. The structure of the inorganic macrounit (see
Figure 4) is, to the best of our knowledge, observed for the first
time, and it can be considered as a new archetype of Zr
octahedra and P tetrahedra assembly.
D
dx.doi.org/10.1021/cg301042y | Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
Figure 5. TG curves of compounds 1 and 2.
5.3%, obs. 5.4%). Beyond this temperature a broad weight loss
is observed (calcd 51.3%, obs. 50.3%) due to the combustion of
the organic part. We can assume that at the end of the analysis
(1200 °C) an equimolar mixture of ZrO2 and ZrP2O7 is formed
because the P/Zr molar ratio is 1 in both compounds.
The thermal behavior of 2 is very similar to that of 1. The
only difference that can be found is the lack of the weight loss
around 280 °C, due to the HF in 1. This behavior is in a very
good agreement with the structural features described above.
(Weight loss: calcd 45.9%, obs 45.3%.)
Structure of 1a. The plateau observed in the TG curve of 1
at 280 °C induced us to study what happened around that
temperature.
Figure 3. Polyhedral representation of the structure of 2 viewed down
the c-axis (a), and a detail of a single layer (b). ZrO4F2 octahedra are
represented in purple, and PO3C tetrahedra are represented in green.
As remarked above, this weight loss can be ascribed to the
evaporation of one HF molecule per formula unit. The
transition is irreversible and yields the stable crystalline
compound 1a. This result was also confirmed by the ion
chromatography analysis on the two solids, which displayed a
decrease of the F/Zr ratio from 3 (in 1) to 2 (in 1a). The high
quality of the diffraction pattern of 1a allowed us to solve the
structure. The structure of 1a is shown in Figure 6.
The loss of the HF molecule left a coordination vacancy in
the zirconium octahedron, and therefore it was saturated by the
neighboring oxygen atom O4 belonging to a carboxylate group
coming from the adjacent chain. An analogous reaction
occurred in a previously reported zirconium derivative,3b but
the structure of the new phase was not solved, though there
were several instances clearly supporting this behavior. The
reaction is shown in better detail in Figure 7. Remarkably,
similar topotactic transformations on metal carboxyphospho-
nates involving the loss of water molecules were recently
reported by the Demadis-Aranda groups.10
The oxygen atom O4, which replaces F2, forms a new bond
with the Zr center re-establishing the octahedral coordination
(Zr−O4 = 2.08(1) Å). As a consequence, the counterion
balancing the positive charge on the nitrogen atom is
represented by the ZrO4F2 octahedra, bearing a formal negative
charge, as in 2.
This solid state reaction is quite slow (the complete
transformation requires about 24 h) and can also be followed
by looking at the cell parameters changes. The a-axis (parallel
to the chain linkage) changes from 14.24 to 12.84 Å, becoming
very similar to the b-axis (12.86 Å). The new cell found has a
pseudotetragonal symmetry, but a reliable structural model in
this symmetry was not found. The new obtained compound 1a
Figure 4. Details of the chain propagation of the CBUs for compound
2 (a and b) and for compound 1 (c) for comparison. The CBUs are
represented by the light blue polyhedra.
and are stacked with an ABA disposition. Moreover, the layer
surface exposes alternate regions of F atoms and aromatic
groups suggesting a very resistant thermal behavior (as
observed from TG data) and a strong hydrophobic character.
Thermal Behavior. Figure 5 shows the coupled TG curves
of 1 and 2. Both compounds show a remarkable thermal
stability.
1 does not lose weight up to about 280°: above this
temperature a weight loss is observed corresponding to a small
plateau ending around 400 °C. This weight loss can be ascribed
to the evaporation of 1 mol of HF/mol of compound (calcd.
E
dx.doi.org/10.1021/cg301042y | Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
FT-IR and Raman. The FT-IR and Raman spectra are shown
in Figures 9 and 10, respectively.
The general aspect of IR spectra displays the characteristic
absorptions of nicotinic and isonicotinic compounds. A
composite system of bands characterizes the 1300−800 cm−1
absorptions; due to the superposition of phosphate and
Figure 6. Polyhedral representation of compound 1a viewed down the
c-axis (a), and a view of one layer (b). ZrO4F2 octahedra are
represented in purple, and PO3C tetrahedra are represented in green.
Figure 7. Details of the transformation of 1 into 1a.
is also a perfect polymorph of 2: the two compounds have in
fact the same composition and molecular weight.
Spectroscopy. The characterization by means of vibrational
(infrared and Raman) and NMR spectroscopies was mainly
carried out to obtain useful information on the local
environment of the octahedral Zr complexes, especially on
the arrangement of the fluorine atoms. This also gave us the
possibility to describe in fine detail the mechanism of the solid
state reaction occurring in 1 to yield 1a. Figure 8 reports the
structure of the octahedral complexes in each compound: the
fluorine atoms were labeled as Fα, Fβ, and Fγ to allow a clear
and unambiguous discussion of all the reported data.
Figure 9. The IR spectra of 1 (a), 1a (b), and 2 (c). For each
compound an inset showing the region between 500 and 600 cm−1 is
included.
Figure 8. Schematic representation of the octahedral complexes
present in 1, 1a, and 2.
F
dx.doi.org/10.1021/cg301042y | Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
1270 cm−1, characteristic of carboxylic acids, are observed in
the IR spectrum of 1. In 1a, a clear indication of unidentate
carboxylate binding to Zr atom comes from the shift of both
CO and C−O vibrations to 1671 and 1341 cm−1,
respectively. A similar effect is also evidenced by the Raman
spectra of these compounds thus confirming the results of IR
detection. In 2, the vibrational pattern of the carboxylate ligand
is at 1660 and 1370 cm−1: again, these signals are both IR and
Raman active. The reduced separation between stretching
bands of double and single CO bond probably suggests that a
higher strength of interaction with Zr atom is established in 2
with respect to 1a. Indeed, a stronger delocalization of the
electronic charge on the metal atom can cause a weakening of
the double bond together with a strengthening of the single CO
bond: this would result in a red-shift and a blue-shift of relevant
stretching signals, respectively.
The low frequency region of Raman spectrum is particularly
efficient at revealing the number and type of Zr−F bonds, and
also the point symmetry of the metal coordination structure. In
fact, according to the mer-ZrO3F3 octahedral structure (point
symmetry C2v), the scattering profile of 1 shows the 551.5
cm−1, 530 cm−1, and 504.3 cm−1 bands, assigned to the
stretchings of the three Zr−F bonds (see Table 2).13 The 504.3
Table 2. Carboxylate and Zr−F Stretching Modes: IR and
a
Raman Frequencies
compound
vibration
IR (cm−1
)
Raman (Δν/cm−1
)
1
ν (CO)
1736.4
1269.9
1734.3
1269.2
551.5
ν (C−O)
b
ν s (FαZrFβ)
ν as (FαZrFβ)
ν (Zr − Fγ)
ν (CO)
(557.5)
(530.0)
(498.3)
1672.6
1341.7
(518.0)
(536.0)
1660.4
1373.5
(560.0)
(531.0)
530.0
504.3
1a
2
1669.4
1332.6
ν (C−O)
c
ν s (FαZrFβ)
ν as (FαZrFβ)
ν (CO)
519.0
537.0
1664.6
1374.0
553.0
525.0
ν (C−O)
ν s (FαZrFγ)
ν as (FαZrFγ)
a
ν: stretching mode; ν s: symmetric stretching; ν as: asymmetric
b
stretching; α, β, and γ indexes refer to Figure 9. Due to the
overlapping with Zr−O vibrations,10 O−C−O and O−C−C
deformations of nicotinic and isonicotinic units,11 the assignment of
low frequency IR absorptions is tentatively proposed in brackets after
c
comparison with Raman bands. Following the normal coordinate
assignment of ref 13b, the symmetric stretching frequency of the trans-
difluorine compound is lower than asymmetric counterpart; in the
other two derivatives one has the opposite.
cm−1 signal is due to the vibration involving the central fluorine
atom: it is not observed for 1a. Conversely, in this sample, the
asymmetric and symmetric stretching bands of trans ligands are
evidenced at 537 and 519 cm−1 respectively. According to a D4h
symmetry of the ZrO4F2 unit, only the symmetric vibration
should be Raman active. Actually, this was the case of the highly
symmetric zirconium derivative that we recently obtained and
characterized (ZPIP in ref 3f): a single band at 544 cm−1 was
observed and assigned to the trans FαZrFβ symmetric
stretching. However, due to the presence of the carboxylate
group in cis position with respect to F atoms, a symmetry
lowering (C2v) can be assumed for 1a, thus allowing both
symmetric and asymmetric modes to be detected with the
Figure 10. The Raman spectra of 1 (a), 1a (b), and 2 (c). For each
compound an inset showing the region between 500 and 600 cm−1 is
included.
pyridine modes of vibration, an intense profile is observed in
this spectral region.3f,11,12 Except for this range and despite the
variety of signals, some clear indications are derived for the
coordination of the zirconium atom. Together with aromatic
and aliphatic CH stretching absorptions, the 3100−2800 cm−1
spectral region highlights the presence of the intense signal of
HF ligand for 1 (absorption at 2942 cm−1 in Figure 9a): this
band is not revealed for 1a and 2, thus confirming the loss of
HF ligand. The CO and C−O stretching modes at 1736 and
G
dx.doi.org/10.1021/cg301042y | Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
absorption and the scattering techniques. Two Raman bands at
547 and 525 cm−1 are also measured for the nicotinic
derivative. As well as the previous case, the symmetric and
asymmetric stretching of Zr−F bonds are both Raman and IR
active due to the low symmetry of the metal coordination
structure (Cs), and a 20 cm−1 separation is measured between
the two spectral components. Notwithstanding these similar-
ities, a different position of Raman bands suggests different sites
of F ligands in the coordination shell of Zr. In fact, the results
obtained for 2 could support the assignment to a cis
arrangement of fluorine ligands, in agreement with the situation
already observed for Pt−F compounds.14
Solid-State NMR. 1H and 19F MAS spectra were recorded at
quite high spinning speeds.
In particular, 1H MAS spectra, reported in Figure 11, show a
very poor spectral resolution, indicating that MAS can only
Figure 12. The 19F MAS spectra of 1 (a), 1a (b), and 2 (c). The
isotropic peaks are marked with stars. The spectra were recorded at
MAS spinning frequencies of 17 and 20 kHz, as indicated in the figure.
nuclei. In order to identify isotropic peaks, distinguishing them
from spinning sidebands, we recorded many spectra at different
MAS frequencies (in the figure only those at 17 and 20 kHz are
reported). Indeed, while isotropic peaks do not change their
resonance position with the MAS frequency, spinning side-
bands do. It is possible to observe from the spectra that 2, 1,
and 2 isotropic peaks can be identified for 1, 1a, and 2,
respectively. In particular, in 1 the two isotropic peaks resonate
at about 3 and 44 ppm: by taking into account the integrals of
isotropic peaks and all sidebands it is possible to estimate a 1:2
ratio between the two groups of signals, indicating that one of
the three fluorine atoms in the molecular structure is
responsible for the signal with an isotropic chemical shift of 3
ppm, while that at 44 ppm arises from two fluorine atoms. On
the basis of molecular structure and symmetry the assignment is
quite straightforward, the two signals corresponding to the
equatorial (Fγ) and axial (Fα and Fβ) fluorine atoms,
respectively. In 1a there is only one distinguishable group of
peaks, with an isotropic chemical shift of about 54 ppm. On one
side, this strongly supports the loss of the equatorial fluorine
atom (as HF) following heating of 1, and on the other side it
reveals a non-negligible change of chemical environment for the
axial fluorine atoms as a consequence of the structural
rearrangement. In 2, two groups of peaks slightly differing for
their isotropic chemical shift (55 and 62 ppm) are observed.
Again, taking into account the integrals of the isotropic peaks
and all the sidebands, the ratio between the corresponding
fluorine atoms is 1:1, although the two groups of signals exhibit
clearly different linewidths. These two groups of signals can in
principle be assigned to either the two fluorine atoms in the
molecular structure (which are indeed slightly inequivalent
because of geometrical distortions from the perfect octahedral
structure) or the two different molecules in the unit cell, as
revealed by the X-ray analysis. Although this doubt cannot be
unequivocally solved at the current stage, by looking at samples
1 and 1a, where small geometrical distortions do not give rise
to a detectable NMR inequivalence, we believe that in 2 the
two signals should be ascribed to the two distinct molecules in
the unit cell. It must be noticed that these two signals show a
clearly different line width, which could arise from different
residual 19F−1H dipolar couplings, probably resulting from a
different exposure of the two fluorine atom pairs to the
neighboring organic chains.
1
Figure 11. The H MAS spectra of 1 (a), 1a (b), and 2 (c). The
spectra were recorded at a MAS spinning frequency of 22 kHz.
slightly reduce the strong 1H−1H dipolar interactions, typical of
a very rigid organic structure. By looking closely at the spectra,
it is possible to see that the spectra of 1a and 2 are very similar,
and, as far as the “resolved” part of the spectrum is concerned,
they consist of a superposition of some broad peaks in the
aliphatic and aromatic regions. On the contrary, the spectrum
of 1 looks quite different, and it clearly shows a peak at about
12 ppm, which can be straightforwardly assigned to the acidic
proton of the HF group, and which is not present in the spectra
of 1a and 2. This observation is in full agreement with the
hypothesized structures for 1, 1a, and 2, especially with the loss
of a molecule of HF on heating sample 1 to give 1a. The
detailed interpretation of the rest of the spectrum is not trivial
considering the very poor spectral resolution.
19F MAS spectra, reported in Figure 12, show the typical
patterns arising by the presence of many spinning sidebands. In
this case, contrary to 1H MAS spectra, we could achieve a good
spectral resolution since 19F nuclei are affected by weaker
homonuclear (19F−19F) and heteronuclear (19F−1H) dipolar
interactions than 1H nuclei. The main cause of line broadening
is reasonably 19F chemical shift anisotropy, which, due to its
inhomogeneous character, is strongly affected by MAS, which
gives rise to quite narrow isotropic peaks and spinning
sidebands. Furthermore, possible residual homo- and hetero-
nuclear dipolar coupling should also have a main inhomoge-
neous character, as arising from two-body couplings, consid-
ering the absence of an extended network of tightly coupled
H
dx.doi.org/10.1021/cg301042y | Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
(d) Vermeulen, L. A.; Snover, J. L.; Sapochak, L. S.; Thompson, M.
E. J. Am. Chem. Soc. 1993, 115, 11767−11774. (e) Byrd, H.; Clearfield,
A.; Poojary, D. M.; Reis, K. P.; Thompson, M. E. Chem. Mater. 1996,
8, 2239−2246. (f) Vermeulen, L. A.; Burgmeyer, S. J. J. Solid State
Chem. 1999, 147, 520−526. (g) Vermeulen, L. A.; Fateen, R. Z.;
Robinson, P. D. Inorg. Chem. 2002, 41, 2310−2312.
(5) Werner, P. E.; Eriksson, L.; Westdhal, M. J. Appl. Crystallogr.
1985, 18, 367−370.
CONCLUSIONS
■
The information provided by the joint use of PXRD, FT-IR,
Raman, and solid-state NMR allowed us to draw a clear picture
of the phase transition occurring in 1, involving the loss of HF
and the consequent rearrangement of the structure to fill the
coordination vacancy created on the zirconium atom. The
result was the covalent joining of the 1D chains constituting 1
to yield the 2D hybrid layers of 1a. The latter is a polymorph of
2, which is a 2D compound too, only based on a geometrical
isomer of the building block present in 1a. Notably, 2
represents a new structural archetype.
Besides the good results obtained, this work represents a first
example of a thorough study of zirconium phosphonates via
solid-state NMR, so we are planning to extend this kind of
characterization to other compounds of the same class. At the
same time, vibrational spectroscopy was also confirmed to be a
powerful tool for the description of some structural details, as
we already showed in a previous work.
(6) Laugier, J.; Bochu, B. LMGP-Suite; ENSP/Laboratoire des
Materiaux et du Genie Physique: 38042 Saint Martin d’Heres, France,
2002.
(7) Favre-Nicolin, V.; Cerny, R. J. Appl. Crystallogr. 2002, 35, 734−
743.
(8) Falcioni, M.; Deem, M. W. J. Chem. Phys. 1999, 110, 1754−1767.
(9) Larson, C.; von Dreele, R. B. Generalized Crystal Structure Analysis
System; Los Alamos National Laboratory, New Mexico, 2001.
(10) (a) Demadis, K. D.; Papadaki, M.; Aranda, M. A. G.; Cabeza, A.;
Olivera-Pastor, P.; Sanakis, Y. Cryst. Growth Des. 2010, 10, 357−364.
(b) Colodrero, R. M. P.; Olivera-Pastor, P.; Cabeza, A.; Papadaki, M.;
Demadis, K. D.; Aranda, M. A. G. Inorg. Chem. 2010, 49, 761−768.
(c) Demadis, K. D.; Famelis, N.; Cabeza, A.; Aranda, M. A. G;
Colodrero, R. M. P.; Infantes-Molina, A. Inorg. Chem. 2012, 51, 7889−
7896.
(11) Stanghellini, P. L.; Boccaleri, E.; Diana, E.; Alberti, G.; Vivani, R.
Inorg. Chem. 2004, 43, 5698−5703.
(12) Socrates, G. Infrared and Raman Characteristic Group Frequencies
− Tables and Charts; John Wiley and Sons Ltd.: Chichester UK, 2001.
(13) Nakamoto, K. Infrared and Raman Spectra of Inorganic and
Coordination Compounds; John Wiley and Sons Ltd.: Hobonek NJ,
2009.
ASSOCIATED CONTENT
* Supporting Information
Crystallographic information file for 1, 1a, and 2 (CIF). Tables
with complete crystallographic parameters for 1, 1a, and 2.
Additional figures. This material is available free of charge via
the Internet at http://pubs.acs.org.
■
S
AUTHOR INFORMATION
Corresponding Author
*E-mail: marcotaddei@hotmail.com.
■
(14) (a) Erlhofer, P.; Preetz, W. Z. Natuforsch. 1989, 44b, 412−418.
̈
(b) Erlhofer, P.; Preetz, W. Z. Natuforsch. 1989, 44b, 1214−1220.
̈
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors thank Dr. Claudia Forte for helpful discussion.
This work was supported by MIUR - Project FIRB 2010 No.
RBFR10CWDA_003.
REFERENCES
■
(1) (a) Clearfield, A. In Progress in Inorganic Chemistry; Karlin, K. D.,
Ed.; John Wiley & Sons: New York, 1998; Vol. 47, pp 374−510;
(b) Vivani, R.; Alberti, G.; Costantino, F.; Nocchetti, M. Microporous
Mesoporous Mater. 2008, 107, 58−70. (c) Vivani, R.; Costantino, F.;
Taddei, M. In Metal Phosphonate Chemistry: From Synthesis to
Applications; Clearfield, A.; Demadis, K., Eds.; Royal Society of
Chemistry: Oxford, 2011; Chapter 2.
(2) (a) Alberti, G.; Casciola, M. Solid State Ionics 2001, 145, 3−16.
(b) Alberti, G.; Casciola, M.; D’Alessandro, E.; Pica, M. J. Mater. Chem.
2004, 14, 1910−1914. (c) Curini, M.; Rosati, O.; Costantino, U. Curr.
Org. Chem. 2004, 8, 591−606. (d) Donnadio, A.; Pica, M.; Taddei, M.;
Vivani, R. J. Mater. Chem. 2012, 22, 5098−5106. (e) Queffelec, C.;
́
Petit, M.; Janvier, P.; Knight, D. A.; Bujoli, B. Chem. Rev. 2012, 112,
3777−3807.
(3) (a) Costantino, U.; Nocchetti, M.; Vivani, R. J. Am. Chem. Soc.
2002, 124, 8428−8434. (b) Vivani, R.; Costantino, U.; Nocchetti, M. J.
Mater. Chem. 2002, 12, 3254−3260. (c) Vivani, R.; Costantino, F.;
Nocchetti, M.; Gatta, G. D. J. Solid State Chem. 2004, 177, 4013−4022.
(d) Vivani, R.; Costantino, F.; Costantino, U.; Nocchetti, M. Inorg.
Chem. 2006, 45, 2388−2390. (e) Taddei, M.; Costantino, F.; Vivani,
R. Inorg. Chem. 2010, 49, 9664−9670. (f) Taddei, M.; Costantino, F.;
Manuali, V.; Vivani, R. Inorg. Chem. 2011, 50, 10835−10843.
(4) (a) Vermeulen, L. A.; Thompson, M. E. Nature 1992, 358, 656−
658. (b) Vermeulen, L. A.; Thompson, M. E. Chem. Mater. 1994, 6,
77−81. (c) Poojary, D. M.; Vermeulen, L. A.; Vicenzi, E.; Clearfield,
A.; Thompson, M. E. Chem. Mater. 1994, 6, 1845−1849.
I
dx.doi.org/10.1021/cg301042y | Cryst. Growth Des. XXXX, XXX, XXX−XXX