Vous êtes sur la page 1sur 10

Article

pubs.acs.org/cm

Rationale of Drug Encapsulation and Release from Biocompatible


Porous Metal−Organic Frameworks
Denise Cunha,† Mouna Ben Yahia,†,‡ Shaun Hall,‡ Stuart R. Miller,† Hubert Chevreau,† Erik Elkaïm,§
Guillaume Maurin,‡ Patricia Horcajada,*,† and Christian Serre*,†

Institut Lavoisier, UMR CNRS 8180, Université de Versailles Saint-Quentin-en-Yvelines, 45 Avenue des Etats-Unis, 78035 Versailles
Cedex, France

Institut Charles Gerhardt Montpellier UMR 5253 CNRS UM2, UM1, Université Montpellier 2, Place E. Bataillon, 34095
Montpellier Cedex 05, France
§
Cristal beamline, Soleil Synchrotron, L’Orme des Merisiers Saint-Aubin, BP 4891192 Gif-sur-Yvette Cedex, France
*
S Supporting Information

ABSTRACT: A joint experimental and computational system-


atic exploration of the driving forces that govern (i)
encapsulation of active ingredients (solvent, starting material
dehydration, drug/material ratio, immersion time, and several
consecutive impregnations) and (i) its kinetics of delivery
(structure, polarity, ...) was performed using a series of porous
biocompatible metal−organic frameworks (MOFs) that bear
different topologies, connectivities, and chemical compositions.
The liporeductor cosmetic caffeine was selected as the active
molecule. Its encapsulation is a challenge for the cosmetic
industry due to its high tendency to crystallize leading to poor loadings (<5 wt %) and uncontrolled releases with a subsequent
low efficiency. It was evidenced that caffeine entrapping reaches exceptional payloads up to 50 wt %, while progressive release of
this cosmetic agent upon immersion in the simulated physiological media (phosphate buffer solution pH = 7.4 or distilled water
pH = 6.3, 37 °C) occurred mainly depending on the degree of MOF stability, caffeine mobility, and MOF−caffeine interactions.
Thus, MIL-100 and UiO-66 appear as very promising carriers for topical administration of caffeine with both spectacular
cosmetic payloads and progressive releases within 24 h.
KEYWORDS: MOFs, encapsulation, release, cosmetic, caffeine

■ INTRODUCTION
Porous metal−organic frameworks (MOFs)1−3 have attracted
loading within a series of functionalized porous flexible
iron(III) terephthalates of MIL-88B topology (MIL = Material
during the past two decades increasing attention due to their of Institut Lavoisier)18 and functionalized rigid zirconium
numerous potential applications2,4−6 in strategic domains such terephthalates UiO-66.19 Interestingly, the most correlated
as gas storage, separation, catalysis, photonics, and, more QSAR descriptors related to caffeine uptake were the polarity,
recently, biomedicine.7−9 Their high and regular porosity polarizability, and H-donor capacity of the grafted functional
combined with their amphiphilic internal microenvironment groups on the organic spacer.
have allowed the achievement of record drug loadings and In this context, there is thus a strong need for a systematic
controlled releases under simulated physiological conditions.10 and rationale study to gain a deeper understanding of the main
This concept has been recently extended to surface-engineered parameters that govern active ingredient encapsulation as well
nanoparticles of nontoxic and biodegradable porous iron(III) as the delivery kinetics from MOFs. As a first step,
polycarboxylates MOFs with very high loadings and progressive encapsulation of the active cosmetic molecule caffeine using a
release of challenging antitumoral or antiretroviral drugs relatively large range of biocompatible porous metal carbox-
coupled with interesting imaging properties.11 Despite this ylates MOFs7,20 with different topologies and compositions
growing interest on drug encapsulation into MOFs12−17 there (structure, porosity, flexible network, functional groups, ...) was
has been no real effort in order to gain understanding of the undergone. Caffeine-containing solids were then characterized
drug adsorption/release processes which makes hazardous using joint experimental and modeling approaches. Finally,
prediction and design of the most suitable MOF for a given their caffeine release kinetics was investigated under two
drug and administration route. To the best of our knowledge,
the only rationale attempt reported so far in this domain was Received: March 11, 2013
realized by some of us dealing with the quantitative structure Revised: June 12, 2013
activity relationship (QSAR) study of experimentally caffeine Published: June 19, 2013

© 2013 American Chemical Society 2767 dx.doi.org/10.1021/cm400798p | Chem. Mater. 2013, 25, 2767−2776
Chemistry of Materials Article

different simulated physiological conditions (phosphate buffer Therefore, better control of caffeine release is strongly desired
solution and water). in order to avoid several topical reapplications and an optimal
To that purpose four different MOF architectures were local concentration during the whole day, with a more effective
selected (Figure 1): (i) the cubic-zeotype mesoporous MIL- slim effect.39,45 Indeed, encapsulation and controlled release of
caffeine still remains a real challenge to be addressed.

■ EXPERIMENTAL SECTION
Synthesis of MOFs. Synthesis of starting materials was performed
under hydrosolvothermal conditions as previously re-
ported.21−23,28,46,47 In a typical synthesis, the metal source and linker
were dispersed in a polar solvent, placed in a Teflon-lined autoclave,
and heated at different temperatures (see details in the Supporting
Information, Table S1). Resulting solids were recovered by filtration
and washed with deionized water and acetone. Then, activation was
carried out to remove the remaining free linker and/or solvent into the
pores by exchange (see details in the Supporting Information; Table
S1).
Caffeine Encapsulation. Caffeine encapsulation was performed
Figure 1. Schematic representation of MIL-100, UiO-66s, MIL-127, by simple impregnation, suspending the (dried or hydrated) MOFs
and MIL-53s structures and caffeine molecule. Iron and zirconium powder into water or ethanol caffeine solutions for rigid and flexible
polyhedra and carbon atoms are in orange, blue, and black, materials, respectively, under stirring at room temperature for 72 h
respectively. Hydrogen atoms have been omitted for a better (for more details see the Supporting Information).
understanding. Caffeine Release. Drug release was carried out by soaking the
drug-loaded powder material in a phosphate buffer solution (PBS; 0.04
100, based on trimers of iron(III) octahedra and trimesate M, pH 7.4) or distilled water at 37 °C under bidimensional continuous
anions leading to a very important porosity (SBET ≈ 2400 stirring (120 rpm, incubator Multiton Orbitale, Infors HT, France). At
m2·g−1, Vp ≈ 1.2 cm3·g−1) associated with two types of different incubation times, an aliquot of supernatant was recovered by
centrifugation and replaced with the same volume of fresh medium at
mesocages (25 and 29 Å) accessible through microporous
37 °C. The amount of encapsulated and released caffeine was
windows (∼4.7*5.5 and 8.6 Å);21 (ii) the cubic structure Soc- determined by combining high-performance liquid chromatography
MOF(Fe) or MIL-12722,23 based on trimers of iron(III) (HPLC), thermogravimetric analysis (TGA), and elemental analysis
octahedra and 3,3′,5,5′-azobenzenetetracarboxylate anions methods (see the Supporting Information).
resulting in two types of pores (an accessible 1D channel X-ray Powder Diffraction (XRPD). XRPD routine patterns were
system (∼6 Å) and cages of around 10 Å accessible through collected in a SIEMENS D5000 diffractometer (θ−2θ) using Cu
windows of ∼4 Å, allowing only adsorption of small molecules Kα1,α2 radiation (λ = 1.54051, 1.54433 Å) from 5° to 15° (2θ) and a
such as H2O or H2 (SBET ≈ 1400 m2·g−1, Vp ≈ 0.7 cm3·g−1)); step size of 0.04° and 4 s per step in continuous mode.
(iii) the cubic UiO-66 solids (UiO for Oslo University), based The XRPD and structure solution for hydrated UiO-66-NH2 was
on zirconium oxoclusters and terephthalate anions bearing carried out using a Bruker D8 powder diffractometer equipped with a
different functional groups on the aromatic benzyl ring (UiO- Lynx-eye detector. Data collection was performed at room temper-
ature in Debye−Scherrer geometry, with the sample in a 0.5 mm glass
66_X for X = H, Br, or NH2); these latter solids possess capillary, with 2θ of 3.75−90°, with a 0.01° step width, and using
octahedral (∼11 Å) and tetrahedral cavities (∼8 Å) accessible monochromated radiation with a wavelength of 1.5409 Å (see the
through microporous triangular windows (∼5−7 Å; SBET ≈ Supporting Information).
1200 m2·g−1, Vp ≈ 0.5 cm3·g−1),24−26 and. finally, (iv) the MIL- The synchrotron powder diffraction experiment has been carried
53 materials built up from corner-sharing chains of iron(III) out for the caffeine-containing MIL-53_Br sample at the Cristal
octahedra related together via functionalized terephthalates beamline, Soleil Synchrotron (Gif-sur-Yvette, France). A monochro-
(MIL-53_X, X = H, Br, or NH2),27−29 creating a three- matic beam was extracted from the U20 undulator beam by means of a
dimensional structure with a microporous 1D pore system; Si(111) double monochromator. Its wavelength of 0.79176 Å was
MIL-53 exhibits a flexible structure30−32 able to reversibly adapt refined from a LaB6 (NIST Standard Reference Material 660a)
its pore size in the presence of different stimuli (temperature, powder diagram recorded just before the experiment. High angular
resolution is obtained, in the diffracted beam, with a 21 perfect crystals
guest adsorption, pressure)33−35 from a closed form (unit cell Si(111) multianalyzer (see the Supporting Information).
volume Vu.c. ≈ 900−1000 Å3; Ø ≈ 4 Å) to an open form (Vu.c. Thermogravimetric Analysis. Samples were analyzed under
≈ 1400−1500 Å3; Ø ≈ 8.5 Å).36 oxygen flow (20 mL·min−1) using a Perkin-Elmer Diamond TGA/
Amphiphilic caffeine (or 1,3,7-trimethylxanthine) was DTA STA 6000 running from room temperature to 600 °C with a
selected as the active molecule due to its small and simple heating rate of 2 °C·min−1.
structure (Figure 1) as well as its large interest for therapeutic Nitrogen Adsorption Porosimetry. N2 isotherms were obtained
and cosmetic applications.37,38 In fact, caffeine acts directly on at 77 K using a Belsorp Mini (Bel, Japan). Prior to analysis,
adipose cells, promoting lipolysis, inhibiting phosphodiesterase, approximately 40−60 mg of activated rigid samples was evacuated for
and increasing cyclic adenosine monophosphate (cAMP) and 12−24 h at 200 °C under vacuum for bare materials or 80 °C for
triglyceride lipase enzyme activity. Caffeine also possesses a caffeine-loaded materials and after caffeine release solids.
Molecular Simulations. A force-field-based Monte Carlo strategy
stimulating effect on cutaneous microcirculation.39 However, was first conducted to provide an estimation of the optimal caffeine
due to its poor skin penetration, caffeine is usually employed in uptake for the different rigid and flexible MOFs, starting with their
cosmetic formulations such as emulsions,40 hydrogels,41 or crystal structures, a microscopic model for the drug, and a suitable
liposomes.42,43 In addition, due to its high tendency to description of the caffeine/MOF interactions including van der Waals
crystallize, the payloads into these formulations are often very and Coulombic contributions. All computational details are reported
poor, not exceding 2−5 wt %, leading to lower efficacies.44 in the Supporting Information.

2768 dx.doi.org/10.1021/cm400798p | Chem. Mater. 2013, 25, 2767−2776


Chemistry of Materials Article

Table 1. Pore Volume and BET Surface Areas before and after Caffeine Encapsulation, Together with the Caffeine Loadings and
Release Times in PBS and Aqueous Media
caffeine release
36
time for total = Kt
prior encapsulation after encapsulation loading capacity release (h) (H2O)
Vp SBET Vp SBET Caf/MOF kinetic
material (cm3·g−1) (m2·g−1) Ø (Å) (cm3·g−1) (m2·g−1) theor (wt %) exp (% wt) (mol/mol) PBS H2O constant
MIL- 0.82 1790 25, 29 0.30 570 65.8 (46.4b) 49.5 ± 1.9 1.70 0.5 48 6.3 ± 0.5
100
UiO-66 H 0.48 1230 7.5, 12 0.22 570 11.7 22.4 ± 3.4 2.14 0.5 48 9.5 ± 0.8
Br 0.30 640 0.11 270 6.8 21.2 ± 0.7 2.20 0.5 48 9.5 ± 0.3
NH2 0.35 930 0.22 580 11.1 13.2 ± 0.2 1.26 2 48 12.5 ± 0.6
MIL-53 H 0.60c 1000c 8.6c 29.5 29.2 ± 1.5 0.36 6 >216a 6.4 ± 0.2
Br 13.1 10.6 ± 3.0 0.19 4 >144a 8.1 ± 0.4
NH2 10.8 0 0.00
MIL- 0.63 1270 10 0.51 1160 31.6 (19.0b) 15.9 ± 0.5 0.63 72 48 13.7 ± 0.5
127
a
Uncompleted release. bSimulated values considering only the accessibility of the large cages (MIL-100) and the 1D channels (MIL-127). cValues
obtained from the MIL-53 solid based on chromium since, contrary to its iron analogue, shows an open porosity when dehydrated.

Periodic density functional theory calculations were performed on caffeine doses with reduced amounts of excipient (in this case,
the MIL-53_X, X = H, Br, and NH2, solids in the absence and the MOF).
presence of caffeine with the aim to (i) determine the magnitude of Analysis of the Drug-Loaded MOFs. Crystalline structure
pore contraction upon encapsulation and (ii) elucidate the geometries
of the drug molecules within the porosity and the resulting interaction
of all solids was checked by XRPD, confirming not only that the
energy. Initial atomic coordinates of the open pore form for each MIL- impregnation step does not alter their crystalline structure but
53 structure were taken from our previous studies,26 and we further also, in accordance with a significant change in the Bragg peaks
considered incorporation of one caffeine per pore in a simulation box relative intensity, the filling of the pores by caffeine molecules
consisting of a (1,1,2) primitive cell. Note that such models probed at (Figure S1, Supporting Information). Remarkably, no recrystal-
least three different starting configurations for the drug and that the lized caffeine is observed whatever the MOF, as confirmed by
optimization always converged toward the same positions. Full XRPD (Figure S1, Supporting Information).
geometry optimizations (atomic positions and unit cell volumes) of
Fourier transform infrared spectroscopy (FTIR) confirmed
these structures were realized using the PBEsol GGA functional48 and
pseudopotentials described by the projector-augmented wave method encapsulation of the caffeine in all loaded solids after
(PAW)49,50 as implemented in the Vienna ab initio simulation package impregnation through the presence of vibrational bands at
(VASP).51−54 A plane wave cutoff of the calculations was set to 600 eV around 1770 and 1658 cm−1 characteristic of the ν(CO)
to ensure convergence, while a k-point mesh of 4 × 4 × 6 was groups of the caffeine (Table S3 and Figure S7, Supporting
considered to represent the Brillouin zone of each system. We Information) as well as a shift to higher wavelengths of these
employed a DFT+U approach as formulated by Dudarev et al.55 to bands in comparison to the pure caffeine, suggesting establish-
overcome the strongly correlated character of the Fe d orbitals. To that
ment of weak interactions between the caffeine moieties and
purpose, Ueff was fixed at 5 as this value was previously shown to
reproduce the structural, electronic, and magnetic properties of the the hybrid framework which induce only a lower electron
MIL-53(Fe) solid.56 Finally, the interaction caffeine/MIL-53_X energy delocalization in the two carbonyl groups. Furthermore, in
was further obtained by the difference between the energy of the similar molecules to caffeine-bearing urea groups (ethyleneurea
caffeine-containing MIL-53_X system and the energy sum of the single and propyleneurea), the negative shift of the CO band has
constituents. previously been associated with interactions involving the

■ RESULTS AND DISCUSSION


Caffeine Encapsulation. Caffeine encapsulation was
oxygen atoms of the carbonyl group, whereas positive shifts
were associated with coordination through the nitrogen
atom.57,58 Therefore, based on the remarked positive shifts of
carried out by a simple impregnation method by suspending this band in the caffeine-containing solids, one can assume that
the powder MOFs into nontoxic and environmentally friendly the N3 atom of the caffeine present between the two carbonyl
aqueous or ethanol caffeine solutions (see Experimental groups (see Figure 1) is involved in the interactions between
Section). Optimal caffeine loadings were reached by suspending caffeine and the MOF materials.
the previously dehydrated powder solids (material:caffeine ratio Force-field-based Monte Carlo simulations were first
= 1:2 and 5:3 for rigid and flexible solids, respectively) into employed to evaluate the theoretical caffeine uptake for each
aqueous (10 mg·mL−1 for rigid MOFs) or ethanol (3 mg·mL−1 investigated MOFs in order to check if the maximum loading
for flexible MIL-53 solids) caffeine solutions at room has been experimentally achieved. Table 1 shows that for the
temperature under magnetic stirring for 72 h (see section flexible MIL-53_X, the simulated values are in very good
below). MIL-100, MIL-53, UiO-66, and MIL-127 materials agreement with the experimental payloads, which emphasizes
showed exceptionally high caffeine payloads (up to 50, 30, 24, that caffeine has been optimally entrapped in the −H and −Br
and 16 wt %, respectively; Table 1). Such caffeine loadings are forms, the predicted uptake for the −NH2 version being similar
among the highest reported so far (2−5 wt %).41 These porous to those obtained for MIL-53_Br. Further, one can observe that
MOFs as well as the previously reported MIL-88B solids (up to the so-obtained caffeine encapsulation in the nonfunctionalized
22 wt %)18 represent new promising candidates for formulation material is very similar to the performance of the Cr analogue,
of caffeine that would allow administration of important ca. 30 wt %.36
2769 dx.doi.org/10.1021/cm400798p | Chem. Mater. 2013, 25, 2767−2776
Chemistry of Materials Article

Regarding MIL-100, the theoretical loading (65.8 wt %) water molecules within a tetrahedral cage, illustrating the
estimated by assuming a full accessibility of both small and large network of hydrogen bondings. This is in agreement with
cages is much higher than the experimental one (49.5 wt %). In interatomic distances between two consecutive water molecules
contrast, if one assumes that the pentagonal windows (∼4.7 × of 2.69 or 3.06 Å, while Owi−Nj distances are 3.06 or 3.47 Å.
5.5 Å) are nonaccessible to caffeine (∼7.6 × 6.1 Å), leading to Thus, any encapsulated drug molecule will have to break these
selective adsorption within only the larger cavities, the intramolecular interactions in order to diffuse within the
predicted (46.4 wt %) and experimental uptakes are in very tetrahedral cages. This explains why caffeine, as not being
good agreement. This hypothesis is supported by a simple capable of providing strong H-donor or -acceptor hydrogen
structural analysis: taking into account the effective volume of a bondings, exhibits a lower loading within the NH2, NO2, or
caffeine molecule (165.5 Å3) and considering the volume of 2OH derivatives of UiO-66.19
each small and large cage (∼8200 and 12 750 Å3) present in a Further, it has been shown by a joint XRPD and DFT
1:2 ratio, the maximum caffeine cargo would be around 1.10 g approach that caffeine preferentially occupies the more
of caffeine per gram of MIL-100. The important discrepancy confined tetrahedral cages,61 leaving empty or only partially
between the theoretical and experimental loading values (1.10 occupied the octahedral cages. Such a statement provides an
vs 0.49 g·g−1) is again consistent with the preferential filling of explanation of the relatively high N2 residual porosity obtained
the larger cages, with a predictable occupation of ca. 77 caffeine after encapsulation (Table 1) and thus suggests that
molecules, which corresponds to ca. 0.48 g·g−1, in accordance optimization of the encapsulation performance could be
with the experimentally observed capacity. Such predictions are achieved at least for the −H and −Br forms that are not
validated by the presence of a significant residual porosity expected to show a high affinity for the solvent.
accessible to nitrogen after caffeine impregnation evidenced Analysis of the Drug Loading Parameters. The
from adsorption isotherm measurements at 77 K on the loaded conditions of caffeine encapsulation were studied by investigat-
sample (see Table 1, pore volume = 0.30 g·cm−3, BET surface ing the influence of different parameters such as the nature of
area = 570 m2·g−1). the solvent, initial state of material hydration, drug/material
The presence of residual pore volume and BET surface area ratio, immersion time, and number of consecutive impregna-
also holds true for MIL-127 and might be associated again to tions.
steric hindrance of caffeine within the porosity of this material Role of Solvent. Nontoxic water and ethanol (rat oral lethal
that would not allow its accommodation within all the dose 50%, LD50 = 10.6 g·kg−1)65 were first selected as solvents
accessible volume for nitrogen. To confirm such findings, because of the high caffeine solubility (∼10 or 3 mg·mL−1,
optimal uptakes have been calculated considering either the full respectively) and their easy removal at low temperatures (<120
accessibility of the cages (∼10 Å) and the 1D channel (∼6 Å) °C). Remarkably, the nature of the solvent used during the
or only the 1D channel assuming that the windows giving impregnation step drastically affects the caffeine encapsulation
access to the cages are too narrow (∼4 Å) to be crossed by rate. For the rigid porous materials based on oxocentered
caffeine. Table 1 shows that the predicted value (19 wt %) trimers of iron(III) octahedra, i.e., MIL-100 and MIL-127
assuming the accommodation of caffeine only the 1D channel is materials, very low capacities (1 and 0 wt %) were obtained
very close to the experimental data (16.5 wt %), which supports when caffeine/ethanol solutions were used in comparison to
only a partial occupancy of the available volume in MIL-127. caffeine aqueous solutions (50 and 16 wt %). This is likely to be
Finally, regarding UiO-66s, except for the amino form, due to the preferential adsorption of ethanol against caffeine for
simulations based on the assumption of a rigid framework such solids. These MOFs exhibit indeed a significant amount of
significantly underestimate the experimental payloads. Such a iron-unsaturated Lewis metal sites (CUS),21,23,66 i.e., up to two
discrepancy is not surprising as it was previously evidenced that CUS per trimer, that could lead to formation of a direct
in such a narrow windows MOF type the dynamics of the coordination adduct of the polar solvent molecules on the iron
phenyl ring strongly affects the adsorption/diffusion processes CUS located on the border of the windows.67 For instance, in
of small gas molecules.59,60 However, these predictions cannot the case of MIL-100, in comparison to water coordination, the
indeed be valuable to check the encapsulation efficiency in the geometry-optimized structures in the presence of ethanol
nonfunctionalized and the −Br forms, one can already argue coordinating the CUS sites show that the free diameters of both
that for the −NH2 version a good agreement between the the pentagonal (1.7 vs 4.5 Å) and the hexagonal windows (4.1
experimental uptake and the simulated one using this vs 8.2 Å; Figures S10 and S11, Supporting Information)
inadequate rigid framework suggests that the encapsulation is considerably decrease (ΔØ ≈ 3−4 Å), leading to more
not optimized at least for UiO-66_NH2. In the later case, the restricted accessibility for the caffeine. Considering the caffeine
steric hindrance of caffeine is most probably induced by dimensions (∼7.6 × 6.1 × 1.8 Å), these poor caffeine uptakes
coadsorption of the solvent, i.e., water, that tends to form using ethanol solutions could therefore be a consequence of the
hydrogen bonding with the amino group that further prevents blocking of both microporous windows. Furthermore, as
the accessibility of the tetrahedral (∼8 Å) and octahedral cages discussed above, one can argue that the active molecule will
(∼11 Å) for caffeine. be preferentially adsorbed only within the larger cages.
In order to gain a deeper understanding of the role of the Further, higher caffeine payloads were achieved in UiO-66
interaction between water and the polar functional group of when using caffeine aqueous solutions in comparison to ethanol
modified UiO-66 solids, the crystal structure of hydrated UiO- ones (22.4 vs 2.5 wt %). This solid also possesses zirconium
66_NH2 was solved from XRPD data (see Figure S2, CUS,25 able to coordinate polar solvents. Similarly to MIL-100
Supporting Information). It appears that within the tetrahedral and MIL-127 solids, one can speculate that ethanol could be
cages an ordering of the hydrogen bondings occurs highlighting coordinated to the zirconium metal sites, decreasing the
the strong affinity of water molecules for the free amino groups accessibility of the triangular windows. In addition, adsorption
of the framework. Figure S3, Supporting Information, shows of ethanol seems to be energetically more favored than water in
the organization of one of the two independent (Ow1) free iron trimer-based materials (MIL-100 and MIL-127) and UiO-
2770 dx.doi.org/10.1021/cm400798p | Chem. Mater. 2013, 25, 2767−2776
Chemistry of Materials Article

66. Finally, the different caffeine concentration in both solvents MIL-100, suggesting that higher encapsulation rates than 50 wt
might also have an influence (see below part drug:material % as suspected from molecular simulations are feasible.
ratio). However, the poor stability of the caffeine aqueous solution
In contrast, caffeine loading was higher when the flexible at concentrations higher than 10 mg·mL−1 led to an important
MIL-53 was suspended in ethanol caffeine solutions than in inaccuracy in the caffeine payload estimation which rules out
aqueous ones. This observation is consistent with a full pore such encapsulation tests. Nevertheless, to further optimize the
opening of the structure in the presence of ethanol (Vu.c. ≈ caffeine loadings, repeated impregnations were performed for
1500 Å3), whereas a narrow pore form is observed in liquid MIL-53 and MIL-100; however, they did not lead to any
water (Vu.c. ∼ 990 Å3).29,68,28 Therefore, the larger pore size of significant increase in the loading capacity. This indicates that
the ethanol form (∼ 8.5 vs 6.0 Å) would play in favor of the maximum caffeine loadings were reached at the selected
caffeine insertion. Interestingly, a higher encapsulation rate was caffeine concentrations, pointing out that the pores were
achieved in ethanol when starting from the anhydrous instead completely filled or that the caffeine/solvent packing did not
of the hydrated form of MIL-53 (29.3 vs 12.7 wt %). As allow any accommodation of additional caffeine molecules.
recently reported,69 the pore content of the ethanol open form, One can estimate the affinity of each solid by calculating the
when starting from the hydrated solid, consists of a mixture of caffeine partition coefficient between the solvent and the MOF
alcohol and water moieties rather than pure alcohol guests. (or encapsulation efficiency = adsorbed drug/drug amount in
Consequently, the caffeine−matrix interactions might not be the starting solution). At low concentrations (<1.2 g·L−1), more
strong enough to replace ethanol and water guests from the than 90% of the caffeine present in the starting ethanol solution
MIL-53 pores, leading to lower drug entrapping when hydrated is adsorbed on the flexible MIL-53, indicating a strong affinity.
materials are used. Accordingly, the adsorption enthalpy for Lower efficiencies were found for the UiO-66 and MIL-100
ethanol is lower than for water, which might favor replacement solids in the same domain of concentrations (1 g·L−1; 40 and
of ethanol over water.70,71 55%, respectively), suggesting lower affinities. If the efficiency is
In light of such findings, caffeine entrapping was carried out
normalized by the maximum caffeine payloads of both solids,
in aqueous solutions and ethanol ones for the rigid and flexible
the affinity of UiO-66 is twice as big as that of MIL-100 (2.0 vs
materials, respectively, starting from the dehydrated flexible
1.1, respectively). Among others, this could be explained by the
materials.
more hydrophobic environment of UiO-66 and its smaller
Drug:Material Ratio. The active molecule:MOF ratio
parameter was then investigated by modifying the starting cavities, which may lead to a higher confinement of caffeine
solution concentration for 72 h, considering the maximum moieties.
solubility values of caffeine in both solvents (∼10 or 3 mg·mL−1 Kinetics of Encapsulation. The kinetics at which the active
in water or ethanol, respectively). Hence, adsorption isotherms molecule is entrapped within the pores is another interesting
of caffeine onto MIL-100, UiO-66, and MIL-53 solids revealed point to be addressed (see Figure 2 (right)). All solids showed
a loading improvement proportional to the concentration very fast kinetics of encapsulation, with, for instance, caffeine
increase up to a plateau, where an equilibrium was achieved loadings higher than 26, 43, and 9 wt % during the first hour of
(Figure 2 (left) and Table 2). The maximum loadings, attained the immersion for MIL-100, MIL-53, and UiO-66, respectively,
which corresponds to around 89%, 87%, and 40% of their total
caffeine capacities, consistent with a relatively high drug−MOF
affinity. Nevertheless, strong host−guest interactions are usually
associated with a slower diffusion at the molecular level
throughout the pores. Consequently, the fast encapsulation
kinetics of caffeine might indicate indeed a moderate strength
of the host−guest interactions, in concordance with the absence
of highly active functional groups on the caffeine able to
strongly interact with the surface of the hybrid solid.
Note also that UiO-66 exhibits slower kinetics than MIL-100
Figure 2. Room-temperature caffeine adsorption isotherms (left) and and MIL-53 (2 vs 1 h). This fact might not be a consequence of
kinetics of encapsulation (right) for MIL-100, UiO-66, and MIL-53. either the dimensionality of the pore system (interconnected
3D in both MIL-100 and UiO-66 rigid solids, 1D channels in
Table 2. Caffeine Encapsulation Loadings in Different MIL-53) or the pore size (MIL-100, UiO-66, and the open
Hydration State and Solvents form of MIL-53−4.7 × 5.5−8.6, 5−7, and 8.5 Å, respectively)
but more likely to the competitive adsorption of the caffeine/
caffeine encapsulation (wt%)
solvent (ethanol or water) mixture within these solids. Indeed,
material hydration state water (10 g·L−1) ethanol (3 g·L−1) diffusion of water within the slightly more hydrophobic
MIL-100 anhydrous 49.5 ± 1.9 1.2 ± 0.2 character of UiO-66 might be enough to slow down the
MIL-127 anhydrous 15.9 ± 0.5 0 dynamics of the caffeine throughout the pores.
MIL-53 hydrated 3.4 ± 1.2 12.7 ± 4.8 Impact of the Ligand Functionalization. Following our
anhydrous 2.2 ± 0.9 29.3 ± 1.5 previous studies,18,19 which evidenced the key factors that
UiO-66 anhydrous 22.4 ± 3.4 2.5 ± 1.4 govern caffeine uptake into a series of functionalized MIL-88B
and UiO-66 solids, the effect of grafting the organic linker on
at the equilibrium concentration, were 20 and 27 wt % for UiO- the caffeine encapsulation and release performance of MOFs
66 and MIL-53 using initial caffeine solution concentrations of was investigated for both the flexible MIL-53 and the rigid UiO-
10 and 1.6 mg·mL−1, respectively. At the considered caffeine 66 solids bearing either hydrophilic and protic NH2 groups and
aqueous concentrations, the equilibrium was not reached for highly polarizable −Br atoms (Figure 1).
2771 dx.doi.org/10.1021/cm400798p | Chem. Mater. 2013, 25, 2767−2776
Chemistry of Materials Article

First, it appears that the MOF functionalization strongly solid were fully geometry optimized (atomic position and cell
impacts the loading capacity of caffeine (Table 1). The bare parameters) in the absence of caffeine, and their resulting
MIL-53 showed a caffeine capacity almost twice as big as that of crystallographic features were further compared to those
MIL-53_Br (0.36 vs 0.19 mol·mol−1), whereas MIL-53_NH2 previously reported by XRPD.28 As shown in Table S5,
did not show any significant encapsulation rate. On the Supporting Information, very good agreement is obtained
contrary, in comparison with the bare UiO-66 solid, caffeine between the experimental and the simulated unit cell volumes
encapsulation into UiO-66_Br was only slightly higher (2.20 vs (relative error < 0.5%) and one can also notice that the b
2.14 mol·mol−1) but significantly lower into UiO-66_NH2 parameters, which are related to the pore opening of the
(1.16 vs 2.14 mol/mol) (Table 1). Indeed, functionalization structure, are very well reproduced which emphasizes that the
of UiO-66 and MIL-53 can act in two different ways: (i) the method employed (DFT+U, PBE functional, pseudopotential,
steric hindrance of the functional group that reduces the pore ...) is accurate enough to describe such solids. The caffeine-
size and/or pore accessibility; (ii) formation of specific containing MIL-53_Br system was further geometry optimized.
interactions between the adsorbate (active ingredient and This calculation confirmed that drug incorporation induces a
solvent) and the functional group. significant contraction of the pore, the unit cell parameters and
Considering the case of MIL-53, the deviation in the volumes matching remarkably the experimental data mentioned
encapsulation rates can be explained by different factors. First, above (a = 19.749 Å, b = 10.844 Å, c = 6.956 Å, ß = 111.2°, V =
the absence of encapsulation within the amino form is most 1388.7 Å3). As the indexing of the caffeine-containing MIL-53
probably due to a stabilization of the framework in its narrow was not possible due to the presence of a mixture of different
pore form due to establishment of strong intraframework structural phases corresponding to different pore opening, a
interactions between the amino group of the linkers and the plausible structure model was proposed using our DFT
oxygen atoms from the iron octahedral that cannot be approach with the following unit cell parameters (a = 19.395
overcome upon adsorption as previously reported for the Fe Å, b = 10.115 Å, c = 6.878 Å, ß = 107.4°, V = 1289.3 Å3)
analogue,28 which would prevent any caffeine adsorption. corresponding to a more significant contraction of the pore
Second, the presence of bulky and heavy bromine atoms (van than its −Br analogue (19% vs 11%), consistent with what has
der Waals radius ≈ 1.85 Å) within the 1D pore system of MIL- been previously reported upon water adsorption.28 Such a trend
53_Br induces a significant decrease in the pore accessibility can be explained by a steric hindrance caused by the Br atom
and/or rotation ability of the phenyl rings to let the guest that prevents a more pronounced closure of the structure.
molecules diffuse throughout the channels, resulting in a Finally, the hypothetical MIL-53_NH2 structure in the
significant decrease of the caffeine uptake. presence of caffeine has been also predicted resulting in a
Regarding UiO-66, the situation significantly differs. Despite pore contraction very similar to the nonmodified material (a =
the presence of the bulky bromine atoms, these electronegative 19.506 Å, b = 9.965 Å, c = 6.877 Å, ß = 107.15°, V = 1275.4
and polarizable atoms could enable formation of additional Å3), here again similar to its behavior upon immersion.28
electrostatic interactions between the caffeine and the frame- A further step consisted of understanding the origin of such
work, which seems to favor adsorption of caffeine. Similarly to structure contractions by analyzing the caffeine−MOFs
the bromine form, amino−terephthalate groups can a priori interactions in terms of geometries and resulting energies. It
interact with the caffeine moieties. However, the lower has been shown that the drug is not aligned along the direction
encapsulation of caffeine within UiO-66_NH2 could be here of the tunnel but adopts an orientation at a slight angle to the
related to the protic and high hydrophilic character of the channel axis. The most likely interaction for caffeine in the bare
−NH2 group, which can form preferentially hydrogen bonds MIL-53 involves relatively strong hydrogen bonding between
with water molecules (used as solvent for impregnation) the oxygen atom of its carbonyl group (C4; Figure 1) and the
instead of caffeine, as previously evidenced by XRPD, leading to μ2-OH group present at the MOF surface (d(O4caf−Hμ2‑OH) =
a coencapsulation of both water and caffeine, reducing the 1.70 Å). Such predominant interactions seem at first sight
loading of the cosmetic molecule. contradictory with IR observations; however, one should keep
XRPD is a very useful technique to analyze the sorption of in mind that such calculations are conducted in the presence
guest within highly flexible solids such as MIL-53. Here, XRPD only of the drug, while experimentally, the presence of some
first confirms that the crystalline MIL-53_X (X = H, Br, NH2) residual ethanol (coming from the impregnation) coordinated
structures are kept intact after the impregnation step (Figure to the iron would tend to replace the OH groups and thus
S1, Supporting Information). Note that the loaded MIL- preventing O4caf−Hμ2‑OH interactions. In addition, weaker van
53_NH2 XRPD patterns did not show any change, in der Waals or CH−π interactions were found between the other
agreement with the absence of any active molecule within the carbonyl and the methyl groups of the drug molecule and the
pores (0 wt %; Table 1). Interestingly, the position of the main organic/inorganic part of the matrix (Figure 3a). However, one
Bragg peaks after encapsulation in the flexible MIL-53 and observes that the interacting distance between the methyl
MIL-53_Br solids is strongly affected, consistent with a group that bears N1 and the oxygen atom of the framework is
different pore opening due to the caffeine entrapping inside relatively short (2.32 Å), which supports that N1 is indirectly
the pores. Unit cell parameters of the caffeine-loaded MIL- involved in the interactions with the pore wall, as might be
53_Br solids were indexed using Dicvol and its graphical suggested by IR measurements mentioned above. We also
interface Winplotr. A monoclinic cell (C2/c (no. 15) with a = observe a weak interaction between the N1 atom of the caffeine
19.75 Å, b = 10.84 Å, c = 6.96 Å, ß = 114.6°, V = 1388 Å3) was and the organic part (d(N1caf−Horganic) = 2.50 Å; Figure 1). The
estimated (Figure S4, Supporting Information). To confirm resulting binding energy of −62.6 kJ·mol−1 is higher than those
such findings, DFT calculations were then performed on the we previously obtained for ibuprofen in the same MOF (−57.4
bare MIL-53 and MIL-53_X (X = Br, NH2)/caffeine systems. kJ·mol−1). In the MIL-53_Br solid, one observes that while the
As a validation step of the methodology employed (functional, main hydrogen-bonding interactions between the drug and the
pseudopotential, ...), the initial open pore forms of each MIL-53 μ2-OH group remains almost unchanged (d(O4caf−Hμ2‑OH) =
2772 dx.doi.org/10.1021/cm400798p | Chem. Mater. 2013, 25, 2767−2776
Chemistry of Materials Article

Figure 4. Caffeine release from different porous MOFs in PBS (left)


and in deionized aqueous (right) media at 37 °C.

MIL-53_Br, and MIL-53, respectively (Figure 5 and Table 1).


Moreover, except for the MIL-127 and MIL-53 materials, most
of the caffeine (53%, 68%, 79%, 93%, and 100% for MIL-53_Br,
Figure 3. DFT-optimized structures of MIL-53_X (X = H (a), Br (b), UiO-66_NH2, MIL-100, UiO-66, and UiO-66_Br, respectively)
and NH2 (c)) in the presence of caffeine. Characteristic caffeine/MOF was delivered through a burst effect, i.e., within the first 30 min.
interacting distances are reported in Angstroms (in green, the main In order to understand the origin of this fast release,
hydrogen bonding O4caf−Hμ2‑OH; in blue, the additional interactions degradation of the solids on their constitutive metal and
between caffeine and the organic/inorganic part of the MOF; in
purple, the interactions induced by the presence of the grafted
organic linker was also analyzed via quantitative release of the
function). organic linker (HPLC), XRPD, FTIR, and N2 adsorption
porosimetry (see Supporting Information). Note that we
recently demonstrated the lack of toxicity and direct removal
1.75 Å) compared to the nonmodified material, the −Br atom by the urine and feces of both metal and ligand of several iron
grafted on the organic linker leads to additional interactions carboxylates MOFs.20 It was concluded that apart from MIL-
with caffeine via its aromatic hydrogen atom (d(Br−Hcaf) = 127, complete caffeine release is associated with a total
2.70 Å) and its carbonyl oxygen (d(Br−Hcaf) = 3.25 Å) amorphization of the solids (Figure S1, Supporting Informa-
resulting in a higher binding energy of −67.1 kJ·mol−1. Finally, tion) and a significant release of the organic linker. After the
the amino-functionalized MIL-53 also shows very similar first 30 min of the assay, up to 18%, 31%, 22%, 3%, 78%, and
strength of hydrogen-bonding interactions (d(O4caf−Hμ2‑OH) 81% of the linker was released for, respectively, MIL-53, MIL-
= 1.72 Å) with the drug; however, one can notice that the 53_Br, UiO-66_NH2, MIL-100, UiO-66, and UiO-66_Br, in
−NH2 functional group serves as an additional anchoring point agreement with degradation of the framework (Figure S9,
for caffeine leading to relatively strong interaction with the Supporting Information). PBS medium leads to a framework
oxygen atom of the second carbonyl function (d(HNH2−O2caf) collapse of the metal polycarboxylates MOFs due to
= 2.20 Å) as well as the methyl group (d(HNH2−Hcaf) = 2.10 replacement of the carboxylate linkers by phosphate groups
Å), resulting in a significantly higher binding energy of −69.4 present in the PBS solution and/or formation of iron oxide
kJ·mol−1. Assuming that the initial stage of the delivery process considering the pKa of the carboxylic functions (∼4−5)
occurs without any degradation of the MOFs, such an rendering less favorable formation of stable metal carboxylate
enhancement of the drug/matrix interactions when the solid bonds at several pH units above their pKa. Accordingly, FTIR
is functionalized would be expected to lead to an increase of the (Figure S7, Supporting Information) revealed the presence of
energy barrier required to be passed prior to triggering the ν(PO4−3) bands characteristic of the phosphate groups at
delivery process, thus leading to a slower initial step of drug around 1000 cm−1, and N2 adsorption measurements showed
release in the case of the amino and bromine versions. no residual porosity after caffeine release for all these solids
Caffeine Release. As a further step, caffeine release was (Figure S6, Supporting Information). Therefore, one can
evaluated at 37 °C using either a simulated serum fluid estimate that under PBS conditions the main parameter driving
(phosphate buffer solution (PBS) 0.04 M at pH = 7.4) or pure drug release is degradation of the MOF. Indeed, the kinetics
water (pH = 6.3), considering that for cosmetics (slim and profiles of caffeine delivery and MOF degradation (Figures 5
microcirculation effect) topical applications are envisaged. Note and S9, Supporting Information) almost overlap, confirming
that PBS seems to be a good and simple simulated serum fluid undoubtedly that MOF degradation governs caffeine release,
since similar release kinetics were achieved using other more particularly when the MOF degrades rapidly in PBS.
complex phosphate media (cell culture medium supplemented A more progressive caffeine release to the PBS medium was
with fetal bovine serum).72 The caffeine-containing MOFs were however evidenced from the MIL-53 and MIL-127 solids, with
in both cases soaked in solution under continuous bidimen- complete release after 6 and 72 h. Interestingly, despite the
sional stirring (see Supporting Information) in order to avoid significant decrease of the crystallinity of MIL-53 after 2 h, only
external diffusion phenomena (Figure 5). 5% and 20% of the cosmetic agent were, respectively, delivered
Release in PBS Media. Caffeine delivery was first followed after 30 min and 1 h. The release kinetics from the MIL-53
into PBS (Figure 4). The amount of released caffeine was solid was fitted from 0 to 4 h to a zero-order kinetics ([caffeine]
quantified by HPLC, and the solids were fully characterized = −2.008 + 23.115t; R > 0.99; where [caffeine] corresponds to
after caffeine delivery (see Supporting Information). the concentration of released caffeine and t to the release time),
Except for MIL-127, caffeine release was very fast, in less predictable and independent of the caffeine concentration. As
than 6 h. The total amount of caffeine was released in 0.5, 2, 2, previously shown for the ibuprofen drug,36 this controlled zero-
4, and 6 h from MIL-100, UiO-66, UiO-66_Br, UiO-66_NH2, order release could be related with its hydrophobic 1D pore
2773 dx.doi.org/10.1021/cm400798p | Chem. Mater. 2013, 25, 2767−2776
Chemistry of Materials Article

system, maintaining the same aperture throughout the entire tion, as evidenced by maintenance of porosity comparable to
delivery process. the starting solids (see Figure S6 and Table S2, Supporting
For MIL-127, despite an initial burst effect of ca. 40% Information). Moreover, no release of the ligand to the water
caffeine delivered in the first 30 min, caffeine release of caffeine medium was detected by HPLC under this delivery condition.
was completely achieved after 72 h (Figure 4). This remarkable Caffeine release rates could be classified as follows from the
result might be the consequence of the higher stability of this faster to the slower kinetics: MIL-100 > UiO-66_Br ≈ UiO-66
solid under the PBS conditions, as it has been evidenced not > MIL-127 > UiO-66_NH2 ≫ MIL-53_Br > MIL-53. As one
only by the XRPD (Figure S1, Supporting Information) but could expect, faster release was observed from the larger pores
also by the absence of release of the constitutive ligand (<limit MIL-100 solid, showing however two different delivery steps:
of detection (1.25 mg·L−1), corresponding to 0.08% of the first one characterized by an important “burst” release with
degradation) and the presence of a significant remaining ca. of 60% of the caffeine released in the first 30 min and the
porosity after caffeine release (SBET = 870 m2·g−1, Figure S6 and second one with a progressive release of the remaining 40%
Table S2, Supporting Information). In fact, reduction of the within the following 48 h. Taking into account that most likely
MIL-127 porosity could be associated with partial degradation caffeine is accommodated into the large cages, as previously
and/or the presence of heavier phosphate anions adsorbed on discussed, these two different stages could correspond to the
the porous framework after the delivery tests. Hence, different location of the caffeine within this cavity. During the
progressive caffeine release from MIL-127 solid would be first step (1 h), this would correspond to the departure of the
mostly governed by a controlled diffusion of the caffeine molecules at the center of the cages with no interaction with
through the slightly amphiphilic 1D pores as well as by the surface of the pores. In contrast, the second delivery step
formation of significant host−guest interactions. (1−48 h) would be associated to release of drug molecules
Such a different kinetics of release profile is however only present at the surface of the pore walls interacting through
related to differences in terms of MOF stability in PBS. Indeed, specific interactions (van der Waals and π−π interactions).
the kinetics of degradation of MIL-53 perfectly fits with the Taking into account the surface of one large cage (∼2640 Å2)
caffeine release profile (Figure 5). To explain such a difference and considering that the maximum surface in contact with the
wall might be one-half of the surface of one caffeine molecule
(∼102 Å2), one can estimate that around 34% of the caffeine
(ca. 26 molecules per cage) could be in direct contact with the
wall of the cavity, which is in reasonable agreement with the
second release step of around 40%.
Very slow caffeine delivery rates were observed for the
flexible MIL-53_Br and MIL-53 solids, with a partial release of
Figure 5. Caffeine (black) and organic ligands (red) release from around 70% of the cosmetic within, respectively, 6 and 9 days
different caffeine-containing MOFs under PBS conditions. after the delivery assay. This very long delivery time with a total
progressive release might pave the way for interesting
in terms of stability, one could argue that MIL-53 bears chains applications in the sustained release of active compounds. As
of iron octahedra that are likely to be less hydrolyzed than discussed previously for ibuprofen,36 this result is due to (i) the
trimers of iron octahedra. MIL-127 is nevertheless built up 1D pore system, which restrains diffusion to only one direction,
from molecular building units (trimers) but possesses four (ii) the flexible porosity that adapts the pore size to the guest
carboxylate groups per spacer that increase the complexing molecule, optimizing the confinement effect, reducing the
strength of the linker and thus would make removal of the diffusion rate, and improving the host−guest interactions, and
linker less favorable compared with MOFs built up from di- or (iii) the relative hydrophobic character of the framework, which
tricarboxylates. Note that the highly water stable UiO-66 MOF slows down water diffusion, which favors a slow exchange and
degrades very fast in PBS, probably due to the strong affinity of release of the caffeine. This could also explain why release is
Zr atoms for phosphate groups and/or formation of Zr oxide, faster from MIL-53_Br compared with MIL-53 (6 vs 9 days),
strongly favored out of the very acidic or basic conditions for bromine atoms increasing the hydrophilic character, even if
such high-valence cation. additional bromine caffeine electrostatic interactions and the
Release in Deionized Water. The study of caffeine release steric hindrance of the bromine atom might play in favor of a
using water medium was performed not only to better mimic slower delivery rate.
cutaneous release but also to rule out the influence of the Full caffeine release was achieved after 48 h from MIL-127,
degradation, allowing a better understanding of the effect of UiO-66, and UiO-66_Br solids. Here, considering the micro-
other parameters (polarity, interactions, flexibility, structure, porous character of the solids, unlike for mesoporous MIL-100,
etc.) on delivery of the therapeutic molecule. all caffeine molecules are mainly interacting in some extent with
Remarkably, caffeine delivery profiles were much slower in the MOF framework. Interestingly, the kinetics of caffeine
pure water than under PBS conditions. This is in agreement delivery can here be empirically adjusted with regression factors
with the much higher stability of the MOFs in deionized water > 0.99 to a Higuchi model ([caffeine] = Kt1/2; Table 2).73,74
(Figure 4). XRPD confirmed the stability of all solids, although Therefore, caffeine release is here mainly governed by a
a broadening of the diffraction peaks was observed in the case diffusion process, predictable by the Higuchi equation and
of MIL-53 solids, which might be explained by the presence of dependent on several factors such as the structure (dimension-
a distribution of forms bearing slight differences in terms of ality, interconnectivity, flexibility, pore size) and composition
pore opening and/or to partial degradation after long time (polarity, interactions). Considering that the external diffusion
periods (up to 9 days; Figure S1, Supporting Information). process is avoided by constant stirring during the assays, the
Additionally, after caffeine delivery in water, N2 adsorption diffusion process is only due to drug motion through the
measurements confirmed the absence of any severe degrada- windows of the cages. Caffeine bears no highly reactive
2774 dx.doi.org/10.1021/cm400798p | Chem. Mater. 2013, 25, 2767−2776
Chemistry of Materials Article

functions that might strongly interact with acidic or basic metal (7) Horcajada, P.; Gref, R.; Baati, T.; Allan, P. K.; Maurin, G.;
or functionalized groups. One expects that using active Couvreur, P.; Férey, G.; Morris, R. E.; Serre, C. Chem. Rev. 2012, 112,
molecules bearing stronger sites, longer deliveries and/or 1232.
significant differences in terms of kinetics of delivery would be (8) Della Rocca, J.; Liu, D.; Lin, W. Acc. Chem. Res. 2011, 44, 957.
(9) Imaz, I.; Rubio-Martinez, M.; An, J.; Sole-Font, I.; Rosi, N. L.;
expected from these solids.


Maspoch, D. Chem. Commun. 2011, 47, 7287.
(10) Horcajada, P.; Serre, C.; Vallet-Regí, M.; Sebban, M.; Taulelle,
CONCLUSION F.; Férey, G. Angew. Chem., Int. Ed. 2006, 118, 6120.
The challenging cosmetic caffeine was successfully entrapped (11) Horcajada, P.; Chalati, T.; Serre, C.; Gillet, B.; Sebrie, C.; Baati,
within porous MOFs, achieving exceptional payloads up to 50 T.; Eubank, F.; Heurtaux, D.; Clayette, P.; Kreuz, C.; Chang, J.-S.;
wt % together bearing very fast kinetics of encapsulation. Hwang, Y. K..; Marsaud, V.; Bories, P.-N.; Cynober, L.; Gil, S.; Férey,
Release of caffeine strongly depends on the release media. In G.; Couvreur, P.; Gref, R. Nat. Mater. 2010, 9, 172.
(12) Sun, C.-Y.; Qin, C.; Wang, C.-G.; Su, Z.-M.; Wang, S.; Wang, X.-
serum-simulated physiological media (PBS pH = 7.4, 37 °C),
L.; Yang, G.-S.; Shao K.- Lan, Z.; Y., Q.; Wang, E.-B. Adv. Mater. 2011,
very fast release occurs that is mainly governed by the MOF 23, 5629.
degradation, except when a sufficient chemical stability is (13) An, J.; Geib, S. J.; Rosi, N. L. J. Am. Chem. Soc. 2009, 131, 8376.
reached. Noteworthy, when other topical simulated conditions (14) Ananthoji, R.; Eubank, J. F.; Nouar, F.; Mouttaki, H.; Eddaoudi,
(distilled water pH = 6.3, 37 °C) are evaluated that do not M.; Harmon, J. P. J. Mater. Chem. 2011, 21, 9587.
degrade the MOFs architectures, progressive releases can be (15) Rieter, W. J.; Pott, K. M.; Taylor, K. M. L.; Lin, W. J. Am. Chem.
achieved that are mainly governed by the degree of caffeine Soc. 2008, 130, 11584.
mobility and MOF−caffeine interactions. Considering that a (16) Taylor-Pashow, K. M. L.; Rocca, J. D.; Xie, Z.; Tran, S.; Lin, W.
typical cosmetic administration typically requires controlled J. Am. Chem. Soc. 2009, 131, 14261.
releases within 8−24 h, MIL-100 and UiO-66 appear thus as (17) Huxford, R. C.; deKrafft, K. E.; Boyle, W. S.; Liu, D.; Lin, W.
very promising carriers for topical administration of caffeine Chem. Sci. 2012, 3, 198.
(18) Gaudin, C.; Cunha, D.; Ivanoff, E.; Horcajada, P.; Chevé, G.;
with both spectacular cosmetic payloads and progressive
Yasri, A.; Loget, O.; Serre, C.; Maurin, G. Microporous Mesoporous
releases. Indeed, slower releases could even be expected Mater. 2012, 157, 124.
under cutaneous conditions since on the upper layer of the (19) Cunha, D.; Gaudin, C.; Colinet, I.; Hall, S.; Horcajada, P.;
skin only partial pressures of vapor water will be usually Maurin, G.; Serre, C. J. Mater. Chem. B 2013, 1 (8), 1101.
reached. Besides, the influence of other components of the (20) Baati, T.; Njim, L.; Neffati, F.; Kerkeni, A.; Bouttemi, M.; Gref,
stratum corneum mainly involving a lipid matrix75−77 and the R.; Najjarm, M. F.; Zakhama, A.; Couvreur, P.; Serre, C.; Horcajada, P.
natural moisturizing factor (aminoacids, lactates, urea, inorganic Chem. Sci. 2013, 4 (4), 1597.
salts, sugars, ...)78,79 could either accelerate or slow down the (21) Horcajada, P.; Surblé, S.; Serre, C.; Hong, D.-Y.; Seo, Y.-K.;
caffeine release rate. Chang, J.-S.; Grenèche, J.-M.; Margiolaki, I.; Férey, G. Chem. Commun.


2007, 2820.
ASSOCIATED CONTENT (22) Liu, Y.; Eubank, J. F.; Cairns, A. J.; Eckert, J.; Kravtsov, V. C.;
Luebke, R.; Eddaoudi, M. Angew. Chem., Int. Ed. 2007, 46, 3278.
*
S Supporting Information (23) Dhakshinamoorthy, A.; Alvaro, M.; Chevreau, H.; Horcajada, P.;
Synthesis, encapsulation, and delivery details, XRD data, FTIR, Devic, T.; Serre, C.; Garcia, H. Catal. Sci. Technol. 2012, 2, 324.
TGA, HPLC, elemental analysis, N2 adsorption porosimetry, (24) Barcia, P. S.; Guimaraes, D.; Mendes, P. A. P.; Silva, J. A. C.;
and simulation methodology. This material is available free of Guillerm, V.; Chevreau, H.; Serre, C.; Rodrigues, A. E. Microporous
charge via the Internet at http://pubs.acs.org. Mesoporous Mater. 2011, 139, 67.


(25) Cavka, J. H.; Jakobsen, S. R.; Olsbye, U.; Guillou, N.; Lamberti,
AUTHOR INFORMATION C.; Bordiga, S.; Lillerud, K. P. J. Am. Chem. Soc. 2008, 130, 13850.
(26) Chavan, S.; Vitillo, J. G.; Uddin, M. J.; Bonino, F.; Lamberti, C.;
Corresponding Author Groppo, E.; Lillerud, K.-P.; Bordiga, S. Chem. Mater. 2010, 22, 4602.
*E-mail: horcajada@chimie.uvsq.fr, serre@chimie.uvsq.fr. (27) Whitfield, T. R.; Wang, X.; Liu, L.; Jacobson, A. J. Solid State Sci.
Notes 2005, 7, 1096.
The authors declare no competing financial interest. (28) Devic, T.; Horcajada, P.; Serre, C.; Salles, F.; Maurin, G.;


Moulin, B.; Heurtaux, D.; Clet, G.; Vimont, A.; Grenèche, J.-M.; Ouay,
B. L.; Moreau, F.; Magnier, E.; Filinchuk, Y.; Marrot, J.; Lavalley, J.-C.;
ACKNOWLEDGMENTS Daturi, M.; Férey, G. J. Am. Chem. Soc. 2009, 132, 1127.
This work was supported by EU funding through the ERC- (29) Millange, F.; Guillou, N.; Walton, R. I.; Grenèche, J.-M.;
2007−209241-BioMOFs ERC (C.S., P.H., D.P., G.M.) and the Margiolaki, I.; Férey, G. Chem. Commun. 2008, 4732.
Region Languedoc Roussillon through the award “Chercheur (30) Férey, G.; Serre, C. Chem. Soc. Rev. 2009, 38, 1380.
d’Avenir” 2009 (G.M.). The authors gratefully acknowledge F. (31) Kitagawa, S.; Uemura, K. Chem. Soc. Rev. 2005, 34, 109.
Ragon and T. Baati for their help in, respectively, material (32) Horike, S.; Shimomura, S.; Kitagawa, S. Nat. Chem. 2009, 1, 695.
(33) Liu, Y.; Her, J.-H.; Dailly, A.; Ramirez-Cuesta, A. J.; Neumann,
preparation and HPLC method development.


D. A.; Brown, C. M. J. Am. Chem. Soc. 2008, 130, 11813.
(34) Beurroies, I.; Boulhout, M.; Llewellyn, P. L.; Kuchta, B.; Férey,
REFERENCES G.; Serre, C.; Denoyel, R. Angew., Chem. Int. Ed. 2010, 49, 7526.
(1) Férey, G. Chem. Soc. Rev. 2008, 37, 191. (35) Yot, P. G.; Ma, Q.; Haines, J.; Yang, Q.; Ghoufi, A.; Devic, T.;
(2) Chem. Soc. Rev 2009, 38, 1201. Serre, C.; Dmitriev, V.; Férey, G.; Zhong, C.; Maurin, G. Chem. Sci.
(3) Chem. Rev. 2012, 112, 673. 2012, 3, 1100.
(4) Acc. Chem. Res. 2005, 215. (36) Horcajada, P.; Serre, C.; Maurin, G.; Ramsahye, N. A.; Balas, F.;
(5) J. Solid State Chem. 2005, 2409. Vallet-Regi, M.; Sebban, M.; Taulelle, F.; Férey, G. J. Am. Chem. Soc.
(6) Férey, G.; Serre, C.; Devic, T.; Maurin, G.; Jobic, H.; Llewellyn, 2008, 130, 6774.
P. L.; De Weireld, G.; Vimont, A.; Daturi, M.; Chang, J.-S. Chem. Soc. (37) Brunyé, T. T.; Mahoney, C. R.; Lieberman, H. R.; Taylor, H. A.
Rev. 2011, 40, 550. Brain Cogn. 2010, 72, 181.

2775 dx.doi.org/10.1021/cm400798p | Chem. Mater. 2013, 25, 2767−2776


Chemistry of Materials Article

(38) Glade, M. J. Nutrition 2010, 26, 932. (71) Bourrelly, S.; Moulin, B.; Rivera, A.; Maurin, G.; Devautour-
(39) Hexsel, D.; Orlandi, C.; Zechmeister do Prado, D. Dermatol. Vinot, S.; Serre, C.; Devic, T.; Horcajada, P.; Vimont, A.; Clet, G.;
Surg. 2005, 31, 866. Daturi, M.; Lavalley, J.-C.; Loera-Serna, S.; Denoyel, R.; Llewellyn, P.
(40) Doucet, O.; Ferrero, L.; Garcia, N.; Zastrow, L. Int. J. Cosmet. L.; Férey, G. J. Am. Chem. Soc. 2010, 132, 9488.
Sci. 1998, 20, 283. (72) Agostoni, V.; Chalati, T.; Horcajada, P.; Willaime, H.; Baati, T.;
(41) Qiu, Y.; Park, K. Adv. Drug Delivery Rev. 2001, 53, 321. Hall, S.; Maurin, G.; Chacun, H.; Bouchemal, K.; Martineau, C.;
(42) Touitou, E.; Levi-Schaffer, F.; Dayan, N.; Alhaique, F.; Riccieri, Taulelle, F.; Couvreur, P.; Rogez-Kreuz, C.; Clayette, P.; Anand, R.;
F. Int. J. Pharm. 1994, 103, 131. Monti, S.; Serre, C.; Gref, R. Adv. Health. Mater. 2013, accepted.
(43) Jeong, S.-Y.; Yi, S. L.; Lim, S.-K.; Park, S.-J.; Jung, J.; Woo, H. (73) Costa, P.; Sousa Lobo, J. M. Eur. J. Pharm. Sci. 2001, 13, 123.
N.; Song, S. Y.; Kim, J.-S.; Lee, J. S.; Lee, J. S.; Park, H. J.; Choi, E. K. (74) Higuchi, W. I. J. Pharm. Sci. 1962, 51, 802.
Int. J. Pharm. 2009, 372, 132. (75) Rawlings, A. V. Int. J. Cosmet. Sci. 2003, 25, 63.
(44) Touitou, E.; Junginger, H. E.; Weiner, N. D.; Nagai, T.; Mezei, (76) Weerheim, A.; Ponec, M. Arch. Dermatol. Rev. 2001, 293, 191.
M. J. Pharm. Sci. 1994, 83, 1189. (77) Bouwstra, J. A.; Gooris, G. S. Open Dermatol. J. 2010, 4, 10.
(45) Bolzinger, M. A.; Briançon, S.; Pelletier, J.; Fessi, H.; Chevalier, (78) Harding, C. R.; Watkinson, A.; Rawlings, A. V. Int. J. Cosmet. Sci.
Y. Eur. J. Pharm. Biopharm. 2008, 68, 446. 2000, 22, 21.
(46) Kandiah, M.; Nilsen, M. H.; Usseglio, S.; Jakobsen, S.; Olsbye, (79) Jacobi, O. Arch. Dermatol. Forsch. 1970, 240, 107.
U.; Tilset, M.; Larabi, C.; Quadrelli, E. A.; Bonino, F.; Lillerud, K. P.
Chem. Mater. 2010, 22, 6632.
(47) Garibay, S. J.; Cohen, S. M. Chem. Commun. 2010, 46, 7700.
(48) Perdew, J. P.; Ruzsinszky, A.; Csonka, G. B. I.; Vydrov, O. A.;
Scuseria, G. E.; Constantin, L. A.; Zhou, X.; Burke, K. Phys. Rev. Lett.
2008, 100, 136406.
(49) Blöchl, P. E. Phys. Rev. B 1994, 50, 17953.
(50) Kresse, G.; Joubert, D. Phys. Rev. B 1999, 59, 1758.
(51) Kresse, G.; Hafner, J. Phys. Rev. B 1993, 47, 558.
(52) Kresse, G.; Hafner, J. Phys. Rev. B 1994, 49, 14251.
(53) Kresse, G.; Furthmüller, J. Phys. Rev. B 1996, 54, 11169.
(54) Kresse, G.; Furthmüller, J. Comput. Mater. Sci. 1996, 6, 15.
(55) Dudarev, S. L.; Botton, G. A.; Savrasov, S. Y.; Humphreys, C. J.;
Sutton, A. P. Phys. Rev. B 1998, 57, 1505.
(56) Combelles, C.; Ben Yahia, M.; Pedesseau, L.; Doublet, M. L. J.
Power Sources 2011, 196, 3426.
(57) Singh, B. R.; Wechter, M. A.; Hu, Y.; Lafontaine, C. Biochem. Ed.
1998, 26, 243.
(58) de Farias, R. F.; da Silva, A. O.; da Silva, U. G., Jr. Thermochim.
Acta 2003, 406, 245.
(59) Yang, Q.; Wiersum, A. D.; Jobic, H.; Guillerm, V.; Serre, C.;
Llewellyn, P. L.; Maurin, G. J. Phys. Chem. C 2011, 115, 13768.
(60) Yang, Q.; Jobic, H.; Salles, F.; Kolokolov, D.; Guillerm, V.;
Serre, C.; Maurin, G. Chem.Eur. J. 2011, 17, 8882.
(61) Devautour-Vinot S., Martineau C., Diaby S., Ben-Yahia M.,
Miller S. R., Serre C., Horcajada P., Cunha D., Taulelle F., Maurin G.
Submitted for publication.
(62) Wiberg, G. S.; Trenholm, H. L.; Coldwell, B. B. Toxicol. Appl.
Pharmacol. 1970, 16, 718.
(63) Yoon, J. W.; Seo, Y.-K.; Hwang, Y. K.; Chang, J.-S.; Leclerc, H.;
Wuttke, S.; Bazin, P.; Vimont, A.; Daturi, M.; Bloch, E.; Llewellyn, P.
L.; Serre, C.; Horcajada, P.; Grenèche, J.-M.; Rodrigues, A. E.; Férey,
G. Angew. Chem., Int. Ed. 2010, 49, 5949.
(64) Leclerc, H.; Vimont, A.; Lavalley, J.-C.; Daturi, M.; Wiersum, A.
D.; Llwellyn, P. L.; Horcajada, P.; Férey, G.; Serre, C. Phys. Chem.
Chem. Phys. 2011, 13, 11748.
(65) Wiberg, G. S.; Trenholm, H. L.; Coldwell, B. B. Toxicology App.
Pharmac. 1970, 16, 718.
(66) Yoon, J. W.; Seo, Y.-K.; Hwang, Y. K.; Chang, J.-S.; Leclerc, H.;
Wuttke, S.; Bazin, P.; Vimont, A.; Daturi, M.; Bloch, E.; Llewellyn, P.
L.; Serre, C.; Horcajada, P.; Grenèche, J.-M.; Rodrigues, A. E.; Férey,
G. Angew. Chem., Int. Ed. 2010, 49, 5949.
(67) Leclerc, H.; Vimont, A.; Lavalley, J.-C.; Daturi, M.; Wiersum, A.
D.; Llwellyn, P. L.; Horcajada, P.; Férey, G.; Serre, C. Phys. Chem.
Chem. Phys. 2011, 13, 11748.
(68) Llewellyn, P. L.; Horcajada, P.; Maurin, G.; Devic, T.;
Rosenbach, N.; Bourrelly, S.; Serre, C.; Vincent, D.; Loera-Serna, S.;
Filinchuk, Y.; Férey, G. J. Am. Chem. Soc. 2009, 131, 13002.
(69) Walton, R. I.; Munn, A. S.; Guillou, N.; Millange, F. Chem.
Eur. J. 2011, 17, 7069.
(70) Devautour-Vinot, S.; Maurin, G.; Henn, F.; Serre, C.; Férey, G.
Phys. Chem. Chem. Phys. 2010, 12, 12478.

2776 dx.doi.org/10.1021/cm400798p | Chem. Mater. 2013, 25, 2767−2776

Vous aimerez peut-être aussi