DOSY-NMR and raman investigations on the self-aggregation and cyclodextrin complexation of vanillin

Article abstract of DOI:10.1021/jp504406j

Vanillin (4-hydroxy-3-methoxybenzaldehyde) is a phenolic aldehyde with limited solubility in water; in this work, we investigate its self-aggregation, as well as its complexation equilibria with ?2-cyclodextrin by using nuclear magnetic resonance (NMR) an

Full text of DOI:10.1021/jp504406j

Article
pubs.acs.org/JPCB
DOSY-NMR and Raman Investigations on the Self-Aggregation and
Cyclodextrin Complexation of Vanillin
Ruggero Ferrazza,† Barbara Rossi,† and Graziano Guella*,†,‡
†Department of Physics, University of Trento, Via Sommarive 14, 38123 Povo (Trento), Italy
‡Biophysical Institute, CNR, Via alla Cascata 56/C, 38123 Povo (Trento), Italy
ABSTRACT: Vanillin (4-hydroxy-3-methoxybenzaldehyde) is a
phenolic aldehyde with limited solubility in water; in this work, we
investigate its self-aggregation, as well as its complexation equilibria
with β-cyclodextrin by using nuclear magnetic resonance (NMR)
and vibrational spectroscopy. In particular, diffusion-ordered NMR
(DOSY) measurements allowing to detect diffusional changes
caused by aggregation/inclusion phenomena lead to a reliable
estimate of the equilibrium constants of these processes, while
Raman spectroscopy was used to further characterize some
structural details of vanillin self-aggregates and inclusion complexes.
Although the self-association binding constant of vanillin in water
was found to be low (Ka ∼10), dimeric species are not negligible
within the investigated range of concentration (3−65 mM); on the
other hand, formation of β-cyclodextrin self-aggregates was not
detected by DOSY measurements on aqueous solutions of β-cyclodextrin at different concentrations (2−12 mM). Finally, the
binding of vanillin with β-cyclodextrin, as measured by the DOSY technique within a narrow range of concentrations (2−15
mM) by assuming the existence of only the monomeric 1:1 vanillin/β-CD complex, was about an order of magnitude higher (Kc
∼ 90) than self-aggregation. However, the value of the equilibrium constant for this complexation was found to be significantly
affected by the analytical concentrations of the host and guest system, thus indicating that Kc is an “apparent” equilibrium
constant.
aromatic interactions essentially consist of van der Waals,
hydrophobic, and electrostatic forces. Other authors have
criticized this point of view and explained the aromatic
interactions just in terms of molecular quadrupoles.1 Even if
the mechanism is still not well understood and no matter what
the right interpretation may be, such interactions drive many
aromatic molecules to form self-aggregates in polar solvents.
Furthermore, it has been seen that several stacking geometries
can occur, such as edge-to-face, face-to-face, and offset stacked,
depending on the substituents at the aromatic ring and on the
solvent wherein these solutes are investigated.4
On one hand, noncovalent interactions can cause self-
aggregation processes, but on the other hand, they are also
responsible for binding phenomena between different mole-
cules as observed in host−guest complexation equilibria where
two interacting species are involved: a big molecule called the
“host”, which usually possesses a central hole, and a smaller
species called the “guest”, that can form the complex by
entering this hole. Among all the available hosts, cyclodextrins
(Figure 1) are probably the most widely used, especially in the
food, cosmetics, and pharmaceutical sectors, as slow-release and
INTRODUCTION
■
Noncovalent interactions are of great interest in physical
chemistry, since they can play many key roles: not only can
they drive molecules to form supramolecular complexes and are
responsible for maintaining the three-dimensional structure of
species such as nucleic acids and proteins, but they are also
relevant in determining phase changes and in molecular
recognition events. Depending on their different nature,
noncovalent interactions are usually classified as electrostatic
(which comprise hydrogen bonds and ionic forces), dispersion,
and hydrophobic interactions. Whatever their origin, they can
affect the chemical and physical properties of the involved
species: the stronger the interaction, the larger these changes.
In any case, the stabilization energy of a molecular cluster is
usually small (between 1 and 100 kJ/mol) and comparable to
the average thermal energy of the kinetic motion of molecules.1
Since this is much smaller than the binding energy of covalent
bonds (about 400 kJ/mol), experimental studies of these
interactions can be very challenging, and only noninvasive
analytical approaches can be reliably applied; basically,
spectroscopic techniques can meet these demands.
When considering aromatic systems, there is no consensus
about what kind of interactions may be dominant, and several
hypotheses have been addressed. For example, Hunter and
Sanders2,3 introduced the “π−π theory”, and supposed that
Received: March 24, 2014
Revised: June 9, 2014
© XXXX American Chemical Society
A
dx.doi.org/10.1021/jp504406j | J. Phys. Chem. B XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry B
Article
in solution, diffusion-ordered NMR spectroscopy (DOSY)
plays a central role, since the measure of diffusion coefficients
of solute species in solution is its primary target.15 In fact,
molecular aggregations cause a change in their diffusion
coefficients,16 according to the Stokes−Einstein equation:
kBT
6πηRH
D =
(1)
where kB is the Boltzmann constant, T the absolute
temperature, η the viscosity of the medium, and RH the
hydrodynamic radius of the species.17 Several NMR pulse
sequences are available for carrying out DOSY experiments, but
all of them rely on the application of magnetic pulsed-field
gradients (PFGs). Basically, in the simplest DOSY-designed
pulse sequence, a magnetization ribbon is first produced by a
90° RF pulse, followed by a PFG which spatially labels the spins
in the sample, generating a magnetization helix; the spins are
then allowed to evolve for a definite time (called the diffusion-
sensitive period). Since their motion partially disrupts this helix,
faster motions lead to more significant losses of the overall
magnetization. At this point, a second PFG (with opposite
polarity to the first) is provided to refocus the magnetization
helix back into a shortened magnetization ribbon, which
generates an attenuated NMR signal. The degree of attenuation
can be analytically related to the self-diffusion coefficient by the
Stejskal−Tanner expression18
Figure 1. Chemical structure and numbering of β-cyclodextrin.
compound-delivery agents.5 They are cyclic oligosaccharides
made up of six (α-CD), seven (β-CD), or eight (γ-CD) D-
glucopyranoside units linked together by 1,4-glycosidic bonds,
and their peculiarity is that in water they assume a torus-like
shape, with a hydrophilic outside and a hydrophobic cavity,
which can therefore admit small nonpolar molecules, increasing
both their aqueous solubility and, eventually, their chemical
stability.6 The process of complex formation in water can take
place only if the size of the guest molecule fits well within the
cavity, which mainly depends on the number of glucose units.
The driving force for the guest inclusion is thought to be the
release of the high-energy water molecules originally present
inside the cavity, and their substitution with the hydrophobic
guest.6
Iq = I0 exp[−Dσ2q2Δ′]
(2)
Vanillin (4-hydroxy-3-methoxybenzaldehyde, Figure 2) is a
natural aromatic compound that can be found in several
where Iq is the measured signal intensity, I0 is the signal
intensity that would be measured in the absence of diffusion, q
= γGδ is a parameter that depends on the gyromagnetic ratio
(γ), the gradient strength (G), and the duration (δ), and Δ′ is
the effective diffusion-sensitive period. Finally, σ is the shape
factor, which accounts for the shape of the applied PFG.
Therefore, to measure diffusion coefficients, a DOSY sequence
is repeated several times, gradually increasing G (and therefore
q), and the obtained intensities are fitted to fulfill eq 2.
Among all its possible applications, the DOSY technique has
also been used to detect aggregation phenomena, although so
far to a limited extent, since interest was mainly focused on
synthetic or natural products mixture analysis. In the present
work, we report a quantitative study of self- and not self-
aggregation (vanillin/β-CD complexation) of vanillin aqueous
solutions. A further structural characterization of these
aggregates was obtained by vibrational spectroscopy through
Raman (and partially IR) measurements carried out in
solutions and/or in solid state preparations.
Figure 2. Chemical structure and numbering of vanillin.
essential oils and in vanilla pods, both in its free form and
bound to larger molecules such as sugar and lignin. Its major
commercial application is as an aroma chemical and, as such, it
has been widely used in the manufacture of chocolate and ice
creams, but also as fragrance in perfumes and cosmetics; it is
estimated that the overall annual production of vanillin exceeds
10 500 tons.7 From an industrial point of view, vanillin can be
obtained by extraction from the pod of the vanilla orchid, but
this is a very expensive operation, so that it only represents 5%
of the vanillin production; the rest is artificially synthesized
from raw materials such as lignin7 and guaiacol.8 Being an
aromatic compound, vanillin is sparingly soluble in water, with
a solubility limit of about 70 mM at room temperature.
Recently, its aqueous self-aggregation has been investigated.9
Also, the ability of cyclodextrins to form complexes with
vanillin has already been addressed by several authors, mainly
with the aim of characterizing the structural property of the
resulting complex.10−14
MATERIALS AND METHODS
■
General Methods. Vanillin (C8H8O3, MW = 152.15 g/
mol) was purchased from Sigma-Aldrich, and β-cyclodextrin
(C42H70O35, MW = 1134.98 g/mol) was obtained from Fluka
Chemie (Switzerland). In order to simplify the features of
NMR and the Raman spectra, its exchangeable protons were
replaced by deuterium atoms by repeated (four times)
dissolving/evaporation cycles in deuterium oxide (D2O,
deuteration degree: 99.97%, Euriso-top, France), followed by
evaporation of the solvent until dryness. Tetradeuterated
ammonium formate (ND4HCOO) was obtained by dissolv-
ing/evaporation cycles in D2O of commercial ammonium
formate (NH4HCOO, Fluka Chemie, Switzerland). A 30 μL
portion of a mother solution (1.0 M, pH 7.0) of ND4HCOO in
Among the developing techniques that allow one to
noninvasively detect aggregation and complexation phenomena
B
dx.doi.org/10.1021/jp504406j | J. Phys. Chem. B XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry B
Article
D2O was used to prepare the buffered NMR solutions (600 μL)
at the final buffer concentration (50 mM).
microprobe setup (Horiba-Jobin-Yvon LabRam HR800)
consisting of an 80 cm focal length spectrograph using a
1800 grooves/mm grating and a charge-coupled device (CCD)
detector cryogenically cooled by liquid nitrogen. The elastically
scattered radiation was filtered by using a notch filter. In this
configuration, the resolution was about 0.28 cm−1/pixel.
The band deconvolution of the Raman spectra in the 1550−
1700 cm−1 wavenumber region was undertaken by using
second derivative computations for evaluating the wave-
numbers of the maxima of the different sub-bands. Multiple
curve fitting into Lorentzian functions was then applied to the
experimental profiles based on these wavenumber values. For
each fitting session, multiple iterations were performed until a
converging solution was reached by minimizing, in the
meanwhile, the value of chi-square.
Low-wavenumber Raman spectra were recorded over the
wavenumbers ranging from 3 to 100 cm−1 by using a triple-
monochromator spectrometer (Horiba-Jobin Yvon, model
T64000) set in double-subtractive/single configuration and
equipped with holographic grating 1800 grooves/mm. Micro-
Raman spectra were excited by the 647.1 nm wavelength of an
argon/krypton ion laser and detected by a CCD detector
cryogenically cooled by liquid nitrogen. Exciting radiation was
focused onto the sample surface with a spot size of about 1 μm2
through a 80× objective with NA = 0.75. The resolution was
about 0.36 cm−1/pixel.
For the vanillin self-aggregation studies, a stock 65 mM
aqueous solution was first prepared, and in order to ensure a
complete solubilization, it was heated up to about 50 °C,
strongly mixed, and sonicated for 15 min. Several (11) aqueous
solutions with concentrations ranging from 3 to 65 mM were
thereby obtained. Measurements on ammonium formate
buffered solutions were carried out on five aqueous solutions
with concentrations ranging from 0.5 to 35 mM. Four aqueous
solutions of β-CD (unbuffered) were prepared in a similar
fashion, with a concentration range from 2 to 12 mM.
For the host/guest complexation studies, a stock aqueous
solution containing both vanillin and β-CD at the 15 mM
concentration was first prepared following the same procedure
described above. From this, another three solutions were
prepared by dilution, obtaining therefore four vanillin/β-CD
1:1 solutions with concentrations ranging from 3 to 15 mM,
which were used to assess the degree of aggregation and to
evaluate the complexation constant.
NMR Measurements. The NMR measurements were
performed with a Bruker-Avance 400 MHz spectrometer
operating with a stationary magnetic field of strength 9.4 T
and equipped with a 5 mm BBI probe capable of providing
longitudinal pulsed-field gradients of up to 53 G cm−1. The 90°
proton pulse length was calibrated and established to be 9.4 μs,
with a transmission power of 0 dB. In all of the measurements,
the temperature was kept at 300.2 K; the chemical shift scale
was referenced to the TSP resonance, which was set at 0.000
ppm. Diffusion coefficients were obtained through DOSY
NMR experiments using the Bruker pulse sequence dstegp3s
The IR spectra were collected in an ATR geometry using an
FT-IR Equinox 55 Bruker spectrometer, at room temperature
with a scanning speed of 30 cm−1 min−1 and a spectral
resolution of 2 cm−1. A few microliters of the solutions (either
vanillin or the vanillin/β-CD inclusion complex) were
deposited on a ZnSe crystal and dried by nitrogen flushing.
The data were analyzed with the software OPUS 2.0.5.
1
(2D H DOSY double stimulated-echo) to compensate for
possible effects of convective motions. The longitudinal eddy
current delay and the gradient recovery delay were kept at fixed
values of 5 and 0.2 ms, respectively. The strength of the pulsed-
field gradients was incremented from 5 to 95% with respect to
the maximum available in 10 increments on a quadratic scale,
whereas the diffusion-sensitive period (Δ) and the gradient
duration (δ) were optimized to allow the signals of interest to
decrease by a factor of 10−20, in order to better characterize
the signal exponential decay predicted by the Stejskal−Tanner
expression (eq 2). The data from the NMR experiments were
processed with the software MestReNova (Mestrelab Research
S.L., Santiago de Compostela, Spain), and diffusion coefficients
were then obtained by integrating several 1H-signals and fitting
the corresponding peaks area to fulfill the Stejskal−Tanner
expression, using the software OriginPro 8 (OriginLab
Corporation, Northampton MA, USA). The D value for a
given solution was taken as the average of the values obtained
by fitting the exponential decays of the several different
resonances (five for vanillin, six for β-CD), and its uncertainty
was taken as the standard deviation of the mean.
RESULTS AND DISCUSSION
Self-Aggregation. NMR Measurements. Before using the
DOSY technique, we proved the existence of vanillin self-
aggregation in water by recording the H NMR spectra of
■
1
several vanillin/water solutions at different concentrations.
After the assignment of vanillin protons (numbers in Figure 2)
for the most diluted solution [3 mM in D2O, 3.979 (OCH3, s),
7.112 (H-5, d, 8.0 Hz), 7.575 (H-2, d, 1.8 Hz), 7.595 (H-6, dd,
1.8 Hz, 8.0 Hz), 9.754 (COH, s)], a significant upfield effect
was observed by increasing concentration, as shown in Figure 3
for H-2, H-5, and H-6. It is worth mentioning that these
changes are not caused by pH changes, eventually induced by
concentration-dependent acid/base species distribution by the
presence of a slightly acidic phenolic group in vanillin. In fact,
NMR measurements carried out on buffered vanillin solutions
(50 mM, ammonium formate) at different vanillin concen-
Vibrational Spectroscopy. All Raman measurements were
carried out both on liquid and dried samples deposited on a
glass slide in air and at room temperature. The corresponding
spectra were obtained in backscattering geometry and by using
two different experimental setups to better explore different
spectral ranges.
The spectra in the wavenumber range between 1550 and
1700 cm−1 were collected by using an exciting radiation at
632.8 nm (He−Ne laser, power at the output ≈20 mW). The
laser was focused onto the sample surface with a spot size of
about 1 μm2 through the 80× objective (NA = 0.9) of a
Figure 3. Intermediate δ region of 1H NMR spectra of vanillin
aqueous solutions taken at three different concentrations.
C
dx.doi.org/10.1021/jp504406j | J. Phys. Chem. B XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry B
Article
D1
trations indicated that the upfield changes of the chemical shifts
of all the vanillin protons were essentially the same as those
measured for the unbuffered solutions. This analysis clearly
showed the salting-out effect of vanillin whose solubility was, as
expected, significantly lower in the buffered than in the
unbuffered solutions.
The observed effects are in agreement with anisotropic-
induced ring currents in the aromatic nucleus of vanillin (π
stacking) which become higher at higher concentration, thus
speaking for the presence of true self-aggregation phenomena.
Since the rate of exchange between the free species and the
aggregates is fast with respect to the NMR time scale, the
observed resonances are found by a weighted chemical shift
average of free species and aggregates:
D =
j
j1/3
(5)
Since the stacking process is usually fast with respect to the
diffusional time scale (Δ ≈ 100 ms), the decaying behavior of
every resonance will be described by a single exponential (see
eq 2), with a measured diffusion coefficient given by the
weighted average
∑ j[Vj]D
j
D1
j
Dmeas
=
=
j2/3[Vj]
∑
[V ]
[V ]
0
0
j
(6)
Within the isodesmic approximation, a single expression can be
obtained in which Dmeas is related to D1, [V0], and Ka:
[V]
j2/3(Ka[V])j−1
∑j j[Vj]δj
Dmeas = D
∑
1 [V ]
δobs
=
0
j
(7)
[V ]
(3)
0
where [V] can be analytically found by solving the equation
[V]
where [V0] is the total vanillin molar concentration (analytical
concentration), [Vj] the concentration of the aggregate
containing a “j” number of vanillin molecules, and δj is the
chemical shift of a given resonance in this aggregate. Actually,
eq 3 provides also a means for quantitatively characterizing the
self-aggregation by an NMR chemical shift analysis, provided
that a proper model is employed for describing the aggregation;
usually, an isodesmic model is assumed,9 which considers the
aggregation process to follow subsequent dynamic binding
steps:
[V ] =
0
(1 − Ka[V])2
(8)
which arises by imposing the mass law conservation.9
With these considerations in mind, we performed several
concentration-dependent DOSY experiments on aqueous
vanillin solutions. In order to ensure that the observed decrease
of the diffusion coefficients was not caused by an increase of the
medium viscosity, we also monitored the diffusion coefficient of
the solvent; the latter was found constant over the whole
investigated vanillin concentration range. In addition, in these
solutions, we did not add buffer or TSP to avoid possible
interferences with the aqueous self-aggregation process.
In Figure 4, the concentration dependence of the measured
D values is shown whereby, relying on an isodesmic model, the
best fitting parameters for Ka and D1 can be evaluated.
V + V ⇄ V
2
V + V ⇄ V
2
3
...
Vj + V ⇄ V
j+1
where each of them is characterized by an aggregation constant,
defined as Ka,j = [Vj+1]/([Vj][V]). In particular, the isodesmic
model assumes that Ka,1 = ... = Ka,j = Ka.
By using only a chemical shift analysis, the dependence of the
chemical shift on the oligomeric state has to be properly
modeled. Usually it is considered to follow the relation9
(j − 1)δmin + δ1
δj =
j
(4)
Here δmin corresponds to the chemical shift of a stack made
up of virtually infinite solute molecules (i.e., high concentration
limit), whereas δ1 is the chemical shift of the considered proton
in the monomer (i.e., infinite dilution limit). With the above-
mentioned assumptions, it is straightforward to express the
observed chemical shift as a function of just Ka, [V0], δ1, and
δmin. Thus, the characterization of aggregation phenomena is
usually carried out by monitoring the chemical shift variations
over a range of solute concentrations, and fitting the values to
fulfill the resulting expression. This procedure has already been
reported for the vanillin aqueous self-aggregation,9 and thus, it
will not be discussed any further.
The DOSY technique relies on a very different physical basis;
in principle, an aggregate of molecules should have a lower
diffusion coefficient than the free molecule itself. As predicted
by eq 1, since the hydrodynamic radius is proportional to the
molecular weight to the 1/3 power, the diffusion coefficient of a
stack of j molecules can be related to that of the free species by
the simple relation 5:
Figure 4. Measured diffusion coefficients as a function of the vanillin
concentration, obtained by DOSY. The solid curve represents the
result of the fitting procedure according to the isodesmic model.
The resulting parameters were D1 = (8.227 0.008) × 10−10
m2 s−1 and Ka = 9.8 0.7 M−1 (χr2 = 1.8). The value of the
measured aggregation constant allows one to draw simple
considerations about the distribution of vanillin among any
potential aggregate species at any given concentration. First of
all, since this value is low, the self-aggregation basically involves
simple oligomers (dimers essentially). A simple calculation for
the 15 mM solution, assuming an isodesmic model, indicates
D
dx.doi.org/10.1021/jp504406j | J. Phys. Chem. B XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry B
Article
that the molar fraction of the free species is significantly higher
(∼88%) than that of dimeric species (∼10%), whereas trimeric
(and higher order) species are almost negligible, accounting
only for about 1% molar fraction.
Things do not significantly change by following a semi-
isodesmic model, where the first aggregation constant (Ka,1,
responsible for the dimerization step) is allowed to differ from
all the others (Ka), assumed again to all be equal. In fact, the
obtained values (Ka,1 = 8.7 0.6 M−1, Ka = 11.7 0.9 M−1 (χr2
= 1.4)) are basically the same as that obtained within the
isodesmic approximation. Worth of note, our results are also in
good agreement with those previously obtained through the
1
analysis of the dependence of H chemical shifts on vanillin
concentration.9
Finally, the measurements of the diffusion coefficients of
vanillin in a buffered solution (pH 7, 50 mM aqueous
ammonium formate) afforded a little bit different D values,
but their dependence on the analytical concentration of vanillin
(i.e., the corresponding equilibrium of self-aggregation) was
found to be almost identical to that measured for the
unbuffered solutions. Thus, the buffer was demonstrated to
not play any relevant role on the aggregation phenomena; it
eventually affects (but only marginally) the solvent viscosity.
Following the characterization of the vanillin self-aggregation,
we used the DOSY technique to investigate the presence of
self-aggregates also in β-CD aqueous solutions. As a matter of
fact, the self-aggregation processes of cyclodextrins have already
been reported.19 In particular, dynamic and static light
scattering measurements, as well as transmission electron
microscopy at cryogenic temperature20 seem to indicate that β-
CD, at concentrations higher than 3 mM, is present in aqueous
solutions as large-size aggregates (diameter ∼100 nm). On the
other hand, previously reported DOSY measurements21 only
indicate a very low decrease in the measured diffusion
coefficients, possibly caused by either a slight increase of the
medium viscosity at higher cyclodextrin concentration or the
formation of a very low number of small aggregates. The results
of our measurements on four aqueous solutions of β-CD at
different concentrations, summarized in Table 1, are in
Figure 5. Raman spectra of (a) vanillin in solid state and (b) a 15 mM
vanillin aqueous solution in the wavenumber range 1550−1700 cm−1.
The experimental data (empty circles) are reported together with the
best-fit (gray line) and the deconvolution components. Inset: FTIR-
ATR spectrum of solid-state vanillin in the same wavenumber range
(1550−1700 cm−1).
involving the stretching vibrations of the carbonyl group of
aldehyde and of the aromatic double bonds of vanillin occur.
On the basis of previous investigations,22,23 these motions are
also expected to be particularly sensitive to the molecular
aggregation processes.
Significant differences in the spectral shape of the vibrational
profile of vanillin in the solid state can be observed with respect
to its spectrum in aqueous solution. For a better inspection of
these spectral differences, the experimental spectra were
deconvolved into their single components, by using a proper
fitting procedure of the data, as shown in Figure 5 (see the
Materials and Methods section for details). The spectrum of
Figure 5a indicates the presence of three distinct components at
1587, 1595, and 1603 cm−1 which are related to vibrational
modes involving the CC bonds of the aromatic ring, while
the peaks falling at 1663, 1669, and 1675 cm−1 can be
essentially assigned to the stretching modes of the CO group
of vanillin. It is noteworthy that the same complex pattern can
be recognized also in the IR spectrum of solid-state vanillin in
the wavenumber range 1550−1700 cm−1 (see inset of Figure
5a), despite the general wider broadening observed for the IR
with respect to the Raman peaks.
Table 1. Concentration Dependence of the Measured
Diffusion Coefficients of β-CD Aqueous Solutions
[β-CD] (mM)
Dmeas (m2 s−1
)
2.1
5.2
(3.09 0.09) × 10−10
(3.10 0.04) × 10−10
(3.09 0.04) × 10−10
(3.01 0.03) × 10−10
The Raman spectrum of the aqueous solution of vanillin
(Figure 5b) shows two peaks at about 1585 and 1598 cm−1
(CC stretching of the aromatic ring) and only one broad
signal at 1669 cm−1 (CO stretching). These assignments are
derived by comparison of experimental data with the results of
DFT calculations carried out on a single molecule of vanillin in
the gas phase and recently reported in the literature.24 On the
other hand, previous X-ray crystallography25 investigations
performed on crystals of vanillin showed that the molecule
crystallizes in a monoclinic crystal system with four vanillin in
the asymmetric unit forming two pairs that lay on two parallel
planes. Each pair of vanillin molecules is supposed to be held
together by hydrogen bonds (HBs) established between the
hydrogen atom of OH of one molecule and the oxygen atom
of the CO group of the other.
8.3
12.5
agreement with this report.21 The results do not suggest the
presence of important β-CD self-aggregation processes, at least
within the sensitivity of the DOSY technique. In principle,
however, the reported heavy aggregates could be present in
such a small amount that they do not affect the averaged
measured diffusion coefficient of β-CD.
Raman/IR Measurements. In order to get insight on some
structural aspects of vanillin self-aggregation phenomena,
Raman spectroscopy measurements were carried out. Figure
5 shows the spectral window in the wavenumber range 1550−
1700 cm−1 for the vanillin solid sample (Figure 5a) and for the
15 mM vanillin aqueous solution (Figure 5b). We chose to
focus our attention on the vibrational spectrum of vanillin in
this spectral region where the vibrational normal modes
By comparing frequencies and relative intensities of the
experimental with calculated peaks, it becomes evident that the
vibrational features of the liquid solution of vanillin are very
E
dx.doi.org/10.1021/jp504406j | J. Phys. Chem. B XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry B
Article
similar to those predicted for the monomer.24 This finding
suggests that, even if the aggregation of vanillin is taking place
in water, in a 15 mM aqueous solution, the main species
present is free vanillin, an outcome perfectly consistent with the
results of DOSY measurements.
1
Table 2. H NMR Chemical Shift Assignments for Free
Vanillin, Free β-CD, and Vanillin/β-CD Complex (15 mM
Aqueous Solutions)
vanillin
β-CD
(15 mM)
complex
(15 mM)
change
(ppm)
proton
(15 mM)
This conclusion is further supported by the inspection of the
low-frequency region of the Raman spectrum of vanillin in solid
state (Figure 6). Similarly to what has been observed on other
vanillin
OCH3
H-5
3.959 (s)
7.088 (d)
7.529 (d)
7.562 (dd)
9.731 (s)
3.985
7.104
7.490
7.574
9.718
+0.026
+0.016
−0.039
+0.012
−0.013
H-2
H-6
COH
β-CD
3.628 (t)
H-4
H-2
H-5
H-6
H-3
H-1
3.618
3.680
3.814
3.886
3.943
5.097
−0.010
−0.015
−0.088
−0.034
−0.066
−0.018
3.695 (dd)
3.902 (m)
3.920 (m)
4.009 (t)
5.115 (d)
Figure 6. Experimental Raman spectrum of solid-state vanillin in the
low-frequency wavenumber range 3−100 cm−1.
molecules,22,26 we find several peaks falling in the spectral range
10−45 cm−1 which have been assigned to bending motions of
the hydrogen bond connecting the two vanillin units.22 These
peaks are not present in the computed spectrum of the
monomeric species24 where the lowest frequency mode is
found at about 108 cm−1 attributable to vibration modes of the
methyl group of vanillin. Therefore, through the inspection of
the low-frequency vibrational modes representing collective
motions of the system,26,27 it is possible to gain useful
information about the presence of molecular aggregation
phenomena.
Figure 7. 1H NMR spectra of the 15 mM β-CD solution (top) and of
the inclusion complex (bottom); the asterisked peak is a vanillin
resonance (OCH3).
Vanillin/β-CD Complexation. NMR Measurements. Fol-
lowing a similar approach to the one used for studying vanillin
self-aggregation processes, before subjecting the vanillin/β-CD
solution system to DOSY measurements, we investigated the
changes of the chemical shifts of the protons of guest (vanillin)
and host (β-CD) in the vanillin/β-CD system eventually
deriving from their not-self-aggregation (complexation), as
compared to the shifts in the separated solutions. The details of
the 1H NMR assignments (in ppm) of the 15 mM solutions of
free vanillin, free β-CD, and vanillin/β-CD complex are listed in
Table 2, where the proton numbering is that reported in
Figures 1 and 2.
chemical shifts are plotted with respect to the host:guest ratio,
in a so-called Scatchard plot.16 In this way, three types of
information are usually achieved: (a) the identification of the
most affected nuclei provides a clue as to the geometry of the
complex; (b) the shape of the titration curve gives quantitative
information about the complexation constant; (c) the
stoichiometry of the complex can be obtained by using the
continuous variation method, which results in a Job plot. These
techniques have already been used for the vanillin/β-CD
system. In fact, the complex formation constant has been
determined using the titration method (1.11 × 104
1800
Strong evidence of the host/guest complexation is clearly
provided by significant changes of several resonances in the
aqueous vanillin/β-CD sample as compared to those of free
M−1),10 the continuous variation method (170.2 M−1),11 and
also the fluorescence spectroscopy (270 M−1).12 Even the
stoichiometry of the complexation reaction has already been
determined to be 1:1.11,13 In addition, more recently, the
complexation between vanillin and β-CD has been assessed by
analyzing changes in UV−visible absorption intensity with
increasing β-CD concentration in excess of vanillin.28 Although
the authors did not extract a numerical value for the
complexation constant, this can be guessed from the
experimental data reported in their work,28 resulting in an
estimated value of about 100 M−1.
1
vanillin and β-CD. As shown in Figure 7 for the restricted H
NMR region where only resonances of the β-CD are observed
(with the exception of the methoxy group of vanillin), the
relevant upfield change of the β-CD resonances indicates that a
host/guest complexation equilibrium is occurring. Similarly to
the self-aggregation, the absence of new resonances confirms
that this equilibrium is a process in fast exchange on the NMR
time scale.
In order to quantitatively study the host−guest complex-
ation, an NMR chemical shift analysis can be carried out
through titration experiments where small aliquots of guest are
added to a solution of host at known concentration, and the
The values obtained by different methods and spectroscopic
techniques do not agree very well with each other, suggesting
that difficulties are prone to work in these investigations. In
order to understand where and how the usual assumptions
F
dx.doi.org/10.1021/jp504406j | J. Phys. Chem. B XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry B
Article
Figure 8. 2D DOSY-NMR measurements on the 15 mM aqueous solutions of vanillin, β-CD, and vanillin/β-CD inclusion complex. The horizontal
axis represents chemical shifts, whereas the vertical axis diffusion coefficients; the dark spots are the resonances of the aqueous solution of the
inclusion complex spread in the second dimension according to their measured diffusion coefficient.
break down, we decided to investigate this system through our
optimized DOSY methodology.
The host−guest complexation is a dynamic process that can
be described by
The green and purple solid lines in Figure 8 represent DmGeas
and DHmeas, respectively, whereas the dashed lines are the
diffusion coefficients of vanillin and β-CD, both as 15 mM
solutions. Basically, the extent to which the solid lines in Figure
8 are displaced from their corresponding dashed lines provides
the basis for the quantitative estimate of the complex formation
constant.
(9)
H + G ⇄ HG
where H, G, and HG denote host (β-CD), guest (vanillin), and
inclusion complex, respectively. Due to the fast exchange
between the free and bound species, for each resonance of
either vanillin or β-CD, there will be a single measured diffusion
coefficient (DGmeas and DmHeas), given by a weighted average:
By taking DH = 3.10 × 10−10 m2 s−1 and DG = 7.78 × 10−10
m2 s−1, the use of eqs 10 and 11 leads to a complexation
equilibrium constant of 86
7 M−1, which is in good
agreement with some literature data.11,12,28 However, this value
was found to significantly change when DOSY measurements
were carried out in aqueous solutions of vanillin/β-CD at
different concentrations (see Table 3), thus indicating that we
G
⎧
⎪
⎩
D
= fG DG + (1 − fG )DHG
meas
⎨
H
⎪
Dmeas = f DH + (1 − f )DHG
H
H
(10)
Table 3. Vanillin/β-CD Complexation Constant Estimated
by the DOSY Technique over the Four Solutions Differing in
the Solute Concentrations
where f G and f H represent the molar fractions of free vanillin
and β-CD, respectively. Now, since f G and f H can be written in
terms of the initial concentrations ([G]i and [H]i) and of
complex concentration at the equilibrium ([HG]eq), provided
that DG and DH are known, the system of eq 10 contains only
two unknown parameters: the diffusion coefficient of the
complex (DHG) and its concentration at equilibrium. This
system can therefore be analytically solved, and once [HG]eq is
found, it is possible to determine the complex formation
constant, defined as
[complex] (mM)
Kc (M−1
)
6.4
10.7
15.0
106
142
86
9
9
7
are not dealing with a true equilibrium constant of 1:1
formation of vanillin/β-CD complex but, instead, only with an
apparent (conditional) equilibrium constant.
[HG]eq
Kc =
This behavior may arise from the existence of the vanillin
self-aggregation phenomena but eventually also from formation
of dimeric species such as (vanillin/β-CD)2, still fulfilling the
1:1 stoichiometry that has been firmly established in several
literature reports11,13 for the vanillin/β-CD system. The
existence of these simultaneous aggregation equilibria may
also explain, by the way, the scattered results found in the
literature for the value of the complexation constant of the
vanillin/β-CD system where such phenomena have been
completely neglected. Worth of note, the dependence of this
apparent equilibrium constant by analytical concentrations of
reactants cannot be explained only on the basis of the
aggregation phenomena, since in this case a monotonic trend
should be observed. For instance, the constant should be higher
in diluted solutions where aggregations eventually leading to
(vanillin)2 and (vanillin/β-CD)2 dimeric species must be
negligible.
([H]i − [HG]eq )([G]i − [HG]eq )
(11)
From eq 10, it is clear that, in order to analytically solve the
problem, DG and DH must be known. From our above-reported
experiments on β-CD aggregation, the latter can be assumed to
be constant, since it does not change in a wide range of β-CD
concentrations. On the other hand, this is not true for DG
(vanillin), which was demonstrated by our investigation to
undergo self-aggregation at high concentrations. In estimating
Kc, we therefore chose the DG values that we have previously
obtained from the vanillin solution at that definite concen-
tration. In particular, we extracted these values from the fitted
curve of Figure 4, for concentrations corresponding to those of
the guest in the studied solution.
DOSY experiments were carried out on the previously
prepared aqueous solutions of vanillin/β-CD. The results for
the 15 mM solution are graphically depicted in the 2D DOSY
plot of Figure 8, which also helps understand the basic principle
allowing to estimate complexation constants by the DOSY
approach.
Raman/IR Measurements. In Figure 9 are reported the
Raman spectra of the inclusion complex formed by β-CD with
vanillin in solid state and in aqueous solution, in the
G
dx.doi.org/10.1021/jp504406j | J. Phys. Chem. B XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry B
Article
carbonyl group of vanillin in the complex in comparison with
vanillin alone.
CONCLUSIONS
■
The identification and characterization of phenomena such as
self-aggregations and host−guest complexations require “soft”
analytical techniques, capable of monitoring the dynamic
equilibrium of the system without altering the equilibrium
itself. In this work, we mainly exploited DOSY-NMR and
Raman spectroscopic techniques to establish whether or not
vanillin undergoes aggregation phenomena in aqueous environ-
ments, as well as to study its complexation within the β-CD
hydrophobic cavity.
The DOSY technique could be reliably applied in the
aqueous solutions, and allowed us to quantitatively describe the
ongoing phenomena: for the vanillin, we were able to estimate
its self-aggregation constant, Ka = 9.8 0.7 M−1, implying that
the aqueous aggregates are just small oligomers; on the other
hand, for β-CD aqueous solutions, DOSY did not detect any
aggregation phenomenon.
Figure 9. Raman spectra of the inclusion complex formed by β-CD
with (a) vanillin in solid state and (b) in aqueous solution (15 mM), in
the wavenumber range 1550−1730 cm−1. The experimental data
(empty circles) are reported together with the best-fit (gray line) and
the deconvolution components. Inset: FTIR-ATR spectrum of solid-
state inclusion complex β-CD/vanillin in the wavenumber range
1550−1730 cm−1.
Although we were able to establish a reliable value also for
the vanillin/β-CD complexation constant (KC = 86 7 M−1), it
significantly changed by changing the host/guest initial
concentrations, thus indicating that we were dealing with an
apparent (conditional) constant. The existence of simultaneous
aggregation equilibria is thought to significantly affect the Kc
value. It is worth noting that also the vanillin distribution
among its self-aggregates is, in turn, concentration-dependent.
The Raman spectroscopy was useful for comparing the solid-
state samples (by definition a self-aggregate state) with the
corresponding aqueous solutions (where self-aggregation may
or may not take place). The spectra of vanillin showed peculiar
patterns in good agreement with those expected24 from DFT
computational calculations assuming the vanillin molecules to
be organized in dimers, and were in accordance with the
already proposed vanillin configuration in the crystal state.25 In
the solid-state inclusion complex, the signals of vanillin (in
particular the stretching of the CO group) resulted in
being slightly shifted toward lower frequency and broader than
those of pure vanillin. In addition, we observed the presence of
a new signal at intermediate frequency (1644.4 cm−1),
presumably caused by the complexation event.
wavenumber range 1550−1730 cm−1. Since β-cyclodextrin does
not show any interfering vibrational bands,22,23 the complex-
ation-induced changes in the vanillin spectrum can be easily
revealed.29
By focusing on the Raman signals associated with the
stretching vibration of the CO group of vanillin, remarkable
changes in the spectral shape of these bands can be observed as
a consequence of complexation of vanillin with β-CD, both in
solid state and in aqueous solution. This finding suggests a
great sensitivity of these vibrational modes to the structural
modifications of the hydrogen bond networks in which the
carbonyl groups of vanillin are involved, in turn related to the
aggregation processes occurring among different vanillin
molecules. The same sensitivity cannot be envisaged on the
analysis of the vibrational peaks between 1550 and 1625 cm−1
related to the motions of the CC bonds of the aromatic ring
of vanillin, which show only a slight broadening of the
vibrational signals as a consequence of complexation with β-
CD, both in solid state and in aqueous solution (Figure 9).
More specifically, the Raman peak found at 1669 cm−1 in
solid state complex (Figure 9a) and attributable to the CO
stretching mode of vanillin appears shifted at 1684 cm−1 in
aqueous solution complex (Figure 9b). The shift to higher
frequencies of the CO stretching mode of vanillin gives
evidence of a breakdown of the intra- and/or intermolecular
hydrogen bond (HB) network in which the carbonyl group is
surely involved. This finding is in agreement with other Raman
spectroscopy investigations of CD inclusion complexes.22,23,29
The spectrum of the solid-state inclusion complex of vanillin
with β-CD (Figure 9a) points out the appearance of an
additional feature at about 1645 cm−1 which is well visible also
in the FTIR-ATR spectrum (see inset of Figure 9a) of the
complex, but it is not present in the profile of the aqueous
solution (Figure 9b). This signal might be attributable to the
shift to lower frequency of the intense peak observed at 1663
cm−1 in the solid state spectrum of free vanillin therein due to
the stretching motion of the CO group of the dimer of
vanillin. This trend could be indicative of the formation, in solid
state, of a wider and stronger HB network involving the
AUTHOR INFORMATION
Corresponding Author
■
*E-mail: graziano.guella@unitn.it. Phone: +39-0461-281536.
Fax: +39-0461-281696.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We would like to thank Prof. Gino Mariotto for making the
micro-Raman apparatus available and Dr. Marco Giarola for his
help. We are also thankful to Dr. Damiano Avi for the FTIR
measurements.
REFERENCES
■
(1) Muller-Dethlefs, K.; Hobza, P. Noncovalent Interactions: A
̈
Challenge For Experiment and Theory. Chem. Rev. 2000, 100, 143−
167.
(2) Hunter, C. A.; Sanders, J. K. M. The Nature of π-π Interactions. J.
Am. Chem. Soc. 1990, 112, 5525−5534.
H
dx.doi.org/10.1021/jp504406j | J. Phys. Chem. B XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry B
Article
(3) Hunter, C. A.; Lawson, K. R.; Perkins, J.; Urch, C. J. Aromatic
Interactions. J. Chem. Soc., Perkin Trans. 2001, 2, 651−669.
(4) Waters, M. L. Aromatic Interactions in Model Systems. Curr.
Opin. Chem. Biol. 2002, 6, 736−741.
Stability and NLO Properties by Density Functional Theory
Calculations. Spectrochim. Acta, Part A 2012, 95 (2012), 354−368.
(25) Velavan, R.; Sureshkumar, P.; Sivakumar, K.; Natarajan, S.
Organic Compounds − Vanillin I. Acta Crystallogr. 1995, C51, 1131−
1133.
(26) Crupi, V.; Fontana, A.; Giarola, M.; Guella, G.; Majolino, D.;
Mancini, I.; Mariotto, G.; Paciaroni, A.; Rossi, B.; Venuti, V.
Cyclodextrin-Complexation Effects on the Low-frequency Vibrational
Dynamics of Ibuprofen by Combined Inelastic Light and Neutron
Scattering Measurements. J. Phys. Chem. B 2013, 117 (14), 3917−
3926.
(27) Hedoux, A.; Decroix, A.; Guinet, Y.; Paccou, L.; Derollez, P.;
Descamps, M. Low- and High-Frequency Raman Investigations on
Caffeine: Polymorphism, Disorder and Phase Transformation. J. Phys.
Chem. B 2011, 115, 5746−5753.
(28) Zeng, Z.; Fang, Y.; Ji, H. Side Chain Influencing the Interaction
Between β-Cyclodextrin and Vanillin. Flavour Fragrance J. 2012, 27,
378−385.
(5) Szejtli, J. Introduction and General Overview of Cyclodextrin
Chemistry. Chem. Rev. 1998, 98, 1743−1753.
(6) Martin Del Valle, E. M. Cyclodextrins and Their Uses: a Review.
Process Biochem. 2004, 39, 1033−1046.
(7) FRIDGE (Fund for Research into Industrial Development,
Growth and Equity). Study Into the Establishment of an Aroma and
Fragrance Fine Chemicals Value Chain in South Africa. Part Three:
Aroma Chemicals Derived From Petrochemical Feedstocks. 2004.
Triumph Ventur Capital (Pty) Limited.
(8) Kumar, R.; Sharma, P. K.; Mishra, P. S. A Review on the Vanillin
Derivatives Showing Various Biological Activities. Int. J. PharmTech
Res. 2012, 4 (1), 266−279.
1
(9) Bogdan, M.; Floare, C. G.; Pîrnau
̆
, A. H NMR Investigation of
Self-Association of Vanillin in Aqueous Solution. J. Phys.: Conf. Ser.
2009, 182, 012002.
(29) Rossi, B.; Verrocchio, P.; Viliani, G.; Mancini, I.; Guella, G.;
Rigo, E.; Scarduelli, G.; Mariotto, G. Vibrational Properties of
Ibuprofen-Cyclodextrin Inclusion Complexes Investigated by Raman
Scattering and Numerical Simulation. J. Raman Spectrosc. 2009, 40 (4),
453−458.
(10) Divakar, S. Structure of a β-Cyclodextrin-Vanillin Inclusion
Complex. J. Agric. Food Chem. 1990, 38 (4), 940−944.
̆
(11) Pîrnau, A.; Bogdan, M.; Floare, C. G. NMR Spectroscopic
Characterization of β-Cyclodextrin Inclusion Complex with Vanillin. J.
Phys.: Conf. Ser. 2009, 182, 012013.
(12) Ishikawa, H.; Kuwano, A.; Matsumoto, K. Complexation of
Vanillin and Ethylvanillin with α-, β-, and γ-Cyclodextrin. J. Fac. Agric.,
Kyushu Univ. 2007, 52 (1), 87−90.
(13) kayaci, F.; Uyar, T. Solid Inclusion Complexes of Vanillin with
Cyclodextrins: Their Formation, Characterization, and High-Temper-
ature Stability. J. Agric. Food Chem. 2011, 59, 11772−11778.
(14) Karathanos, V. T.; Mourtzinos, I.; Yannakopoulou, K.;
Andrikopoulos, N. K. Study of the solubility, Antioxidant Activity
and Structure of Inclusion Complex of Vanillin with β-Cyclodextrin.
Food Chem. 2007, 101, 652−658.
(15) Cohen, Y.; Avram, L.; Frish, L. Diffusion NMR Spectroscopy in
Supramolecular and Combinatorial Chemistry: An Old Parameter −
New Insights. Angew. Chem., Int. Ed. 2005, 44, 520−554.
(16) Pessine, F. B. T.; Calderini, A.; Alexandrino, G. L. Review:
Cyclodextrin Inclusion Complexes Probed by NMR Techniques. In
Magnetic Resonance Spectroscopy; Kim, D.-H., Ed.; InTech: Shanghai,
China, 2012. ISBN: 978-953-51-0065-2.
(17) Price, W. S. Pulsed-Field Gradient Nuclear Magnetic Resonance
as a Tool for Studying Translational Diffusion: Part I. Basic Theory.
Concepts Magn. Reson. 1997, 9, 299−336.
(18) Sinnaeve, D. The Stejskal-Tanner Equation Generalized for Any
Gradient Shape − An Overview of Most Pulse Sequences Measuring
Free Diffusion. Concepts Magn. Reson. 2012, 40A (2), 39−65.
(19) Loftsson, T.; Masson, M.; Brewster, M. E. Self-Association of
̀
Cyclodectrins and Cyclodextrin Complexes. J. Pharm. Sci. 2004, 93,
1091−1099.
(20) Bonini, M.; Rossi, S.; Karlsson, G.; Almgren, M.; Lo Nostro, P.;
Baglioni, P. Self-Assembly of β-Cyclodextrin in Water. Part 1: Cryo-
TEM and Dynamic and Static Light Scattering. Langmuir 2006, 22,
1478−1484.
̀
(21) Guerrero-Martìnez, A.; Gonzalez-Gaitano, G.; Vinas, M. H.;
Tardajos, G. Inclusion Complexes between β-Cyclodextrin and a
Gemini Surfactant in Aqueous Solution: An NMR Study. J. Phys.
Chem. B 2006, 110, 13819−13828.
(22) Crupi, V.; Majolino, D.; Paciaroni, A.; Rossi, B.; Stancanelli, R.;
Venuti, V.; Viliani, G. The Effect of Hydrogen Bond on the Vibrational
Dynamics of Genistein Free and Complexed with b-Cyclodextrins. J.
Raman Spectrosc. 2010, 41 (7), 764−770.
(23) Crupi, V.; Majolino, D.; Venuti, V.; Guella, G.; Mancini, I.;
Rossi, B.; Verrocchio, P.; Viliani, G.; Stancanelli, R. Temperature Effect
on the Vibrational Dynamics of Cyclodextrin Inclusion Complexes:
Investigation by FTIR-ATR Spectroscopy and Numerical Simulation.
J. Phys. Chem. A 2010, 114 (25), 6811−6817.
(24) Balachandran, V.; Parimala, K. Vanillin and Isovanillin:
Comparative Vibrational Spectroscopic Studies, Conformational
I
dx.doi.org/10.1021/jp504406j | J. Phys. Chem. B XXXX, XXX, XXX−XXX