Tracking the Structural Dynamics of Hybrid Layered Double

Feb 15, 2011 - Tracking the Structural Dynamics of Hybrid Layered Double Hydroxides ..... of Suc−Zn2Al LDH consisting of 10 × 9 × 1 rhombohedral (...
1 downloads 0 Views 4MB Size
ARTICLE pubs.acs.org/cm

Tracking the Structural Dynamics of Hybrid Layered Double Hydroxides Julien Pisson,† Nicole Morel-Desrosiers,‡ Jean Pierre Morel,‡ Andre de Roy,§ Fabrice Leroux,§ Christine Taviot-Gueho,*,§ and Patrice Malfreyt*,† †

Laboratoire de Thermodynamique et Interactions Moleculaires (CNRS, UMR 6272, LTIM, F-63177 Aubiere), ‡Laboratoire Microorganismes: Genome et Environnement (CNRS, UMR 6023, LMGE, F-63177 Aubiere), and §Laboratoire des Materiaux Inorganiques (CNRS, UMR 6005, LMI, F-63177 Aubiere), Clermont Universite, Universite Blaise Pascal, BP 10448, F-63000 Clermont-Ferrand, France

bS Supporting Information ABSTRACT: The present study aims at investigating the structural and dynamics properties of intercalated tartrate and succinate anions in Zn2Al layered double hydroxides (LDHs). The comparison between these two anions allows the estimation of the impact of the presence of hydroxyl groups in tartrate on the dynamics, orientational, and rotational behaviors of the species in the interlayer domain. A variety of experimental techniques (X-ray diffraction, 1H-13C MAS NMR variable contact time, proton conductivity, microcalorimetry) is used in association with molecular dynamics (MD) simulations. A specific behavior is then established for the tartrate in terms of rotation around the backbone, diffusion, and organization of the water molecules in the interlayer domain. The difference in the enthalpies of exchange between the two LDHs is calculated from MD simulations and compared to that obtained by microcalorimetry experiments. This energy calculation evidences for the existence of anion-anion interactions that are more favorable for tartrate than for succinate. The combination of experimental and theoretical methods performs very well for the structural, energetic, and dynamics characterization of the interlayer species. KEYWORDS: hybrid inorganic/organic materials, layered materials, theory and modeling

’ INTRODUCTION Layered double hydroxides (LDHs) also known as anionic clays correspond to a class of intercalation compounds with positively charged brucite-like hydroxide layers and intercalated exchangeable interlayer anions to maintain charge neutrality. The general formula of xþ (Ax/nn-) 3 mH2O abbreviated the LDH is [MII1-xMIII x (OH)2] II III II hereafter as A-MRM in which M = Mg, Zn, Co, Ni, Mn, etc., MIII = Cr, Fe, V, Co, etc., An- is an inorganic or organic anion of charge n, and R is the MII/MIII molar ratio. The wide range of cationic composition for the hydroxide layers and the ease with which various inorganic and organic anions can be accommodated into the interlayer space of LDHs, bring LDHs at a high level of interest for the development of advanced nanostructured hybrid materials. Research on hybrid inorganic-organic materials has experienced an explosive growth since the 1980s, with the expansion of soft inorganic chemistry processes allowing for the mixing of the organic and inorganic components at the nanometer scale. Hybrid materials properties rely on the chemical nature of the inorganic and organic components and their synergy. The links and interactions existing at the hybrid interface are therefore of paramount importance. According to the classification stated by Sanchez1 for hybrid organicinorganic materials, LDHs belong to Class I hybrids including all systems where only van der Waals, hydrogen bonding, or electrostatic r 2011 American Chemical Society

forces are present between the organic and inorganic components. On the contrary, in Class II hybrids, parts of the inorganic and organic components are linked through strong covalent or iono-covalent bonds. An important and promising field of application of hybrid LDHs is that of optical devices and pigments.2 The literature already offers several examples of composites of organic chromophore-LDH nanostructured assemblies. Human health is another field of increasing interest for the use of bionanohybrids, and in the last years, studies considering the possible use of LDHs as “molecular containers” have been raised including biopharmaceutical objectives (modification of drug liberation or its solubility), pharmacological targets (avoid or diminish side effects), and chemical factors (increasing of the stability).3-8 The last field with advanced LDH applications is that of polymer nanocomposites.9,10 A steady increase of articles related to the use of LDHs as filler in polymers is observed in the literature. LDH-polymer interactions as well as physical confinement are manifested in material key properties, and these systems have been found Received: October 27, 2010 Revised: January 21, 2011 Published: February 15, 2011 1482

dx.doi.org/10.1021/cm1030848 | Chem. Mater. 2011, 23, 1482–1490

Chemistry of Materials suitable for mechanical reinforcement, in permeation and flame retardancy applications. Owing to the low crystallinity of hybrid LDHs, establishing the structure-property relationship for these materials is not an easy task. Indeed, these materials contain many defects attributed to mismatches in the geometry/charge density of the host layers and guest anions preventing an ideal packing of the slabs and also inhibiting crystal growth. For these reasons, the refinement of LDH structures has been rarely achieved, and to the best of our knowledge, no crystal structure of a hybrid-LDH has been reported so far. Yet, the structure-property control is essential if one wants to tailor and fine-tune properties (optical, thermal, chemical) in very broad ranges and to design advanced hybrid LDH for specific applications. The hydrogen-bond network is probably the major factor determining the structure of hybrid LDH. Indeed, hydrogen bonding being directional, it induces specific orientation of interlayer organic anions with respect to the hydroxide layer and surrounding water molecules, then with major consequences in the dynamic of interlayer species and their reactivity. The molecular simulation methods have been successfully applied to investigate the interlayer properties of various LDHs11-26 which are difficult to access experimentally. These atomistic models are well-adapted to study these heterogeneous systems over length- and time-scales that allow the dynamic modeling of the interlayer species. The main purpose of the present paper is to use a combined approach of improved experimental methods (X-ray diffraction, 1 H-13C MAS NMR variable contact time, proton conductivity, and microcalorimetry) and powerful molecular dynamics simulations to extract detailed information on the interlayer structure, hydrogen bonding, and dynamics of two hybrid-LDHs intercalated with tartrate and succinate anions, as model hybrid-LDH structures. Succinate and tartrate anions were chosen as representative anions, displaying simple and similar chains of carbon atoms with hydroxyl groups in the case of tartrate, making these anions more hydrophilic than succinate anions.27 We believe that the results obtained here prevail in hybrid-LDH materials and can be used to better understand the nanostructure of other hybrid-LDHs intercalated with more complex anion structures.

’ EXPERIMENTAL SECTION This section discusses the preparation of Tar-Zn2Al and SuccZn2Al LDHs and the experimental methods. Sample Preparation. The preparation of Zn2Al-LDH samples studied here was performed using the coprecipitation method followed with anion exchange reactions as reported elsewhere.28 The crystallinity of the samples was improved by applying a postsynthesis hydrothermal treatment.28 The chemical analysis of the phases before the hydrothermal treatment was performed by inductively coupled plasma atom emission spectroscopy at the CNRS Center of Chemical Analysis (Vernaison, France). The chemical compositions given in Table S1 (Supporting Information) are in accordance with the formation of LDH phases in all cases. X-ray Diffraction. Powder X-ray diffraction patterns of the samples were collected on a Siemens D-501 diffractometer with Cu KR1R2 radiation in the range 2θ = 2-110 with a step size of 0.03 and counting time of 30 s/ step. Cell parameters and Rietveld refinements were carried out with FULLPROF suite program. The instrumental contribution to peak broadening was determined with a Y2O3 standard. The anisotropic size broadening effect of the peak profile was modeled with spherical harmonics. The Rietveld structure refinement of Tar-Zn2Al was performed considering the R3m space group and taking as initial values, for the atomic positions within the hydroxide layer, those reported for Cl-Zn2Al;25 the atomic positions for

ARTICLE

tartrate anions were obtained by averaging over conformations collected along the evolution time of the MD simulation. 1D Electron Density Map. The electron density maps29 were obtained from   ¥ 2πlz F00l cos FðzÞ ¼ c l¼0



where c is the unit cell parameter, z is the fractional coordinate along the c stacking axis, and F00l are the structure factors of the 00l reflections. Six isolated 00l reflections were used for calculating the electron density distribution of Tar-Zn2Al along the stacking axis c. F00l were derived from their intensities corrected for Lorentz-polarization effects. The signs of the structure factors were directly obtained from the scattering contributions of Zn2Al hydroxide layers, assuming a relatively small contribution of the intercalated molecules. Solid State NMR Spectroscopy. Formula and NMR peak assignments: Tar-Zn2Al = Zn2Al(OH)6(-OO1C2CHOH3CHOH4COO-) 3 nH2O; Suc-Zn2Al = Zn2Al(OH)6(-OO1C2CH23CH24COO-) 3 nH2O. High resolution 13C CP (cross-polarization) MAS (magic angle spinning) NMR spectra were collected at 75.5 MHz on a Bruker 300 instrument operating at 7.04 T. During the MAS conditions, 4 mmdiameter zirconia rotors were spinned at 10 kHz. The 13C CP MAS NMR experiments were made using a high-power decoupling and run using a (π/2) 1H pulse of 9 μs. The recycling time was 5 s. The Hartmann-Hahn transfer condition was satisfied using glycine as a reference: γ1H B1H ¼ γ13C γ13C Further details can be found in the Supporting Information. CP dynamics experiments were performed using a collection of 500 transients to get a proper signal-to-noise response, especially in the low contact time values. Contact times were ranging from 5 μs up to 50 ms. The resonance peak intensity of nC nuclei of interest 1,4C versus 2,3C was arbitrarily taken instead of peak integration due to ill-defined shape especially in the low contact time values. The spin-lattice relaxation time T1 was measured by an inversionrecovery sequence (π-τ-π/2) and fitting the magnetization intensity variation against time τ by the equation M(t) = M0(1 - 2.0 exp(-t/T1)). Conductivity Measurements: Impedance Analysis. Each studied material was first equilibrated for 3 days at 50% relative humidity atmosphere. Appropriate amounts of powders were then pressed under 750 MPa as disks of 1.300 cm diameter in order to obtain an approximately 0.23 cm thickness, with 0.017 cm graphite electrodes copressed on each side of the cylindrical pellets. The thickness of these pellets was measured with a 0.001 cm resolution leading to a geometrical constant close to 0.15 1/cm. In the measuring cell, each sample took place between two polished platinum electrodes, and an airtight environment was provided by an encircling silicon o-ring firmly leaning on these electrodes. The temperature of the cell was controlled by means of a pair of Peletier effect modules. The conductivity measurements were performed by impedance spectroscopy on a Solartron 1174 frequency response analyzer fitted with Kativois impedance adaptation probes. For each sample, the measurement process involves 26 isothermal sets of measurement, first from room temperature to 253 K, then from 253 to 343 K, and last, return to initial temperature. For each isothermal set, a sinusoidal signal U = 30 mV was applied on the sample, between 0.1 Hz and 1.0 MHz, with 10 measurements per decade. Each measured complex impedance value Z = a þ jb with j2 = -1 is thus retrieved as two scalar values a and b, respectively, for the real and imaginary parts.

’ COMPUTATIONAL SECTION Potential Models. The LDH layers and the dicarboxylate anions were modeled using the all-atom (AA) version of the Dreiding force field.30,31 The water molecules were modeled using the TIP3P model.32 The general potential function is given in 1483

dx.doi.org/10.1021/cm1030848 |Chem. Mater. 2011, 23, 1482–1490

Chemistry of Materials

Figure 1. (a) Conformation of succinate and tartrate dianions with the numbering of the carbon atoms. (b) Snapshots of the simulation supercell of Suc-Zn2Al LDH consisting of 10  9  1 rhombohedral (R3m) unit cells. There are three hydroxide layers and three interlayers. Blue octahedrons are Al atoms, brown octahedrons are Zn atoms, and the white sticks are OH groups of the hydroxide layer with the oxygen atom represented in red. In the interlayer, water molecules are represented by blue gray sticks. The structure of an equilibrium configuration with a complete ordering of Al and 90 water molecules per interlayer is shown.

zthe Supporting Information (SI) as well as the different force constants and equilibrium distances and valence and dihedral angles reported in Tables S5 and S6. The geometries of the tartrate and succinate anions (Figure 1a) were optimized by means of HF (6-31G**) quantum chemistry calculations. The atomic charges of the dicarboxylate anions were fitted to reproduce the molecular electrostatic potential created around each anion at the HF level with the 6-31G** basis. We used the CHELPG33 procedure as a grid-based method. The quantum ab initio calculations were carried out using the GAMESS package.34 The partial charges and the nonbonded interactions parameters are given in Table S6. Methodology. The periodic boundary conditions were applied in the three dimensions. The long-range electrostatic interactions were calculated by the Ewald summation technique.35,36 The parameters of this method were R = 0.3208 Å-1 (convergence parameter) within a relative error of 10-6 and kmax = {9  9  8} (the reciprocal space vectors). The equations of motions were integrated using the Verlet Leapfrog algorithm with 1 fs as the time step. The cutoff radius was 10 Å, and the Verlet list sphere radius was fixed to 12 Å.

ARTICLE

The MD simulations were performed at constant composition in the isothermal-isobaric (NpzzT) ensemble. The temperature was fixed at T = 298 K. By imposing the uniaxial pressure component pzz perpendicular to the LDH layers, only the interlayer spacing was allowed to vary and the x and y dimensions of the simulation cell were kept constant. Within this methodology, the MD simulations were performed more strictly at the NpzzT ensemble. pzz was maintained at 1 atm using the Berendsen algorithm37 with coupling constants 0.1 ps (temperature) and 0.5 ps (pressure). A typical run consisted of 50 ps of equilibration followed by a production phase of an additional 70 ps. The structural and thermodynamic properties were calculated over 7000 configurations saved during the acquisition phase. The configurations were generated using the parallel version of the DL_POLY_MD package38 by using up to 16 processors at a time. Description of the Simulation Cell. The hydroxide layers of the simulation cell were built from the three-layer polytype 3R1 with rhombohedral symmetry (space group R3m) often assumed for synthetic LDH materials, starting from the atomic positions of Cl-Zn2Al LDH.39 The simulation supercell consisted of 10  9  1 hexagonal (R3m) unit cells, containing three metal-hydroxide layers. The simulation cell is thus constructed from the stacking of identical brucite layers by translation of (2/3a þ 1/3b) between consecutive layers. With a Zn/Al ratio equal to 2, each hydroxide layer contains 60 Zn2þ and 30 Al3þ. Hence, the composition of the ZnAl LDH supercell is Zn180Al90(OH)540 with 45 dicarboxylate anions (15 per each interlayer). Ninety water molecules are added into the interlayer space to reproduce the experimental interlayer distance. For a Zn/Al ratio equal to 2, the use of the periodic boundary conditions did not allow to completely satisfy both the cation disordering and the trivalent cations avoidance rule40 within the hydroxide layers. Accordingly, in order to study the influence of the cations avoidance rule, we chose to simulate two pseudosystems. In the systems abbreviated as (Suc-Zn2Al)ord or (Tar-Zn2Al)ord, the cation avoidance rule is satisfied with all the Al-Al distances greater than the intermetallic distance. But in turn, it was not possible to avoid in the simulation cell an ordered distribution of the metal cations with six nearest-neighbor Zn atoms around Al atoms while each Zn atom is surrounded by three nearest-neighbor Zn atoms and three nearestneighbor Al atoms. In the systems abbreviated as (Suc-Zn2Al)dis or (Tar-Zn2Al)dis, the distribution of Al atoms is not ordered, but some Al-Al distances may not respect the above distance criterium, i.e, some Al3þ cations may occupy adjacent octahedra. Snapshots of the simulation supercells of the Suc-Zn2Al is shown in Figure 1b. When the distribution of the metal cation is not explicitly mentionned, the properties are calculated from the ordered distribution.

’ RESULTS AND DISCUSSION At the start of the simulation, the center of mass of the dianions and water molecules were placed randomly in the interlayer region on the 18g and 36i equivalent positions of the R3m space groups, respectively. The MD simulations in the NpzzT ensemble show that it is possible to reproduce the experimental interlayer distance of 12.1 Å for both succinate and tartrate containing LDH, by considering 90 water molecules in each interlayer region which is consistent with the number of water molecules determined by chemical analysis, i.e., 3 water molecules per formula unit41 (see Table S1 of the SI). Reproducing the interlayer distance of 12.1 Å requires use of a partial charge of -1.35 e for the oxygen atom of the OH groups of the layers. This value accurately matches that used in the previously reported molecular simulations of the Cl-Zn2Al.25 1484

dx.doi.org/10.1021/cm1030848 |Chem. Mater. 2011, 23, 1482–1490

Chemistry of Materials

ARTICLE

Figure 2a compares the total atomic density profiles along the c stacking axis of the supercell derived from MD simulation and the electron density along the same direction obtained by Fourier transform of the intensity of the 00l basal X-ray diffraction reflections for tartrate containing LDH. The similarity is quite remarkable and supports the structural model provided by MD. The more intense peaks correspond to the atomic/electron densities of hydroxide layers

with three contributions visible on the atomic profiles: cations in the center, and OH groups in the upper and lower surface of the hydroxide layer. The carboxylate groups together with water molecules are responsible for the two other maxima located at the outer parts of the interlayer space; the average distance between these maxima and the nearest OH groups of the adjacent hydroxide layer is about 2 Å which is consistent with hydrogen-bonding interactions. The two other peaks of very low intensity, located in the middle of the gallery, arise from CH-OH groups surrounded by a few water molecules. Figure 2b zooms on the local composition of a specific interlayer region and shows the profiles of water molecules in the interlayer region for the Tar-Zn2Al and Suc-Zn2Al LDHs. We also represent the profiles of the hydrogen atoms of the hydroxide layers to better visualize the interlayer spacing. Figure 2b shows two different locations of water molecules in the interlayer region: one at 4.0 Å from the center of hydroxide layer and the second at the central plane of the interlayer space (at 6.0 Å). The first position corresponds to a region constituted by an important number of water molecules and extends over 2 Å and may correspond to a hydrophilic part of the interlayer region where hydrogen bonds can form between water molecules, carboxylate groups, and OH groups of the hydroxide layer. The formation of a monolayer on each side of the hydroxide layer has been already observed from previous computer simulations of MgAl LDHs with intercalated organic anions.12,22,26 The distribution of water molecules in the central layer is almost uniform in SucZn2Al indicating that this second zone presents no preferential location of the water molecules. On the other hand, the small maximum observed in the case of Tar-Zn2Al indicates a localization of water molecules in the central zone slightly more marked. This slight difference can be ascribed to the presence of the hydroxy groups of the tartrate anions which tends to make this zone more hydrophilic. This smaller proportion of water molecules in the midplane of the interlayer has been shown from MD simulations of MgAl LDHs with different interlacated organic anions.12,22 The tilting of the guest molecules with respect to normal to the hydroxide layer can be calculated from the average angles between the longest axis of the guest molecule and normal to the hydroxide layers. The longest axis of the dianion molecules is calculated from the diagonalization of the inertia tensor (see the Supporting Information). The angle values and the percentages of molecules for which this angle is less than 45 are reported in Table 1 for both an ordered and disordered distribution of cations within the hydroxide layers. The angular distributions are given for completeness in Figure S4 for the two types of Al distributions. Several orientations coexist with however an important statistical weight for the perpendicular orientation. One also can see that the orientation of the intercalated dicarboxylate anions slightly depends on the cation distribution within the hydroxide layers. Indeed, for an ordered distribution, the guest

Figure 2. (a) Total density profiles along the z direction of the supercell simulation (black solid line) and electronic density profile (red dashed line) for tartrate containing LDH. (b) Density profiles of the oxygen atoms of water molecules for Suc-Zn2Al (dashed blue line) and Tar-Zn2Al (solid red line) LDHs. The dashed-dotted line corresponds to the density profile of the hydrogen atoms of the hydroxide layer (peaks around 2 and 10 Å).

Table 1. Average Angle between the Longest Axis of the Dianion and the Hydroxide Layera system

Dx (10-12 m2/s)

Dy (10-12 m2/s)

Dz (10-12 m2s-1)

D (10-12 m2/s)

angle (deg)

N%

(Suc-Zn2Al)ord

18

100

33

48

23

(Suc-Zn2Al)dis

32

71

40

30

9

26

(Tar-Zn2Al)ord

23

90

24

35

16

25

(Tar--Zn2Al)dis

35

67

15

15

9

13

4

4

8

5

2400

2400

2400

2400

(Cl-Zn2Al)ord isotropic bulklike water

35

a N is the percentage of molecules whose tilt angle is less than 45. D is the diffusion coefficient defined by ((Dx þ Dy þ Dz)/3) where DR represents the diffusion coefficient in the R direction. The diffusion coefficient of pure isotropic water calculated from MD simulations is also given for comparison.

1485

dx.doi.org/10.1021/cm1030848 |Chem. Mater. 2011, 23, 1482–1490

Chemistry of Materials

ARTICLE

Figure 3. Contour maps of normalized atomic probability densities in the plane of the interlayer (x,y) of Suc-Zn2Al and Tar-Zn2Al LDHs with 90 intercalated water molecules. (a and c) Oxygen atoms of water molecules (red contours); oxygen atoms of carboxylate groups (yellow contours). (b and d) Hydrogen atoms of the dicarboxylate dianions. The small white points correspond to the positions of the 36i crystallographic sites in the R3m space group.

anions are oriented more perpendicularly with respect to the hydroxide layers. A disordered distribution of Al cations within the layer induces coexisting regions with higher and smaller charge densities. This local heterogeneity in the charge density changes the distribution of the tilt angle (see Figure S4 of the SI). The distribution of the tilt angle is no longer zero for the high values of the tilt angle contrary to the distribution calculated in

the ordered LDH. As a result, the ordering distribution of the Al cations enhances the propensity for the succinate and tartrate anions to be oriented along the direction normal to the layer. Interestingly, the analysis of the atomic density profiles of hydrogen atoms (see Figure 3), in interlayer (x,y) planes, reveals a different dynamical behavior of the interlayer anions with succinate anions that can rotate around the C1-C4 axis whereas 1486

dx.doi.org/10.1021/cm1030848 |Chem. Mater. 2011, 23, 1482–1490

Chemistry of Materials

ARTICLE

Figure 5. Components of the mean squared displacements of intercalated water molecules as a function of time for the Suc-Zn2Al and Cl-Zn2Al LDHs.

Figure 4. CP curves (versus tCP) for 2,3C nuclei in Tar-Zn2Al (filled square) and Suc-Zn2Al (open square) for tCP < 1 ms. The inset shows the same curve over the entire time domain.

tartrate molecules show no rotation. Such contrast in the dynamical behaviors has never been reported so far and will be explained below by the tartrate-tartrate energy contributions. Experimental evidence of these different dynamical behaviors for tartrate and succinate intercalated anions will be provided thereafter by NMR and proton conductivity measurements. Together with the spin-lattice relaxation time, T1, the dynamics of the intercalated anions was probed through the measurement of the 1 H-13C cross-polarized magnetization curve as a function of the contact time (see Figure 4) as exemplified in the review by Kolodziejski and Klinowski.42 T1 was measured equal to T1(2,3C)Suc of 0.45 s, T1(2,3C)Tar of 1.5 s and T1(1,4C)Suc of 0.65 s, T1(1,4C)Tar of >2 s; however a CP contact time experiment will picture a rather more complete description that the straightforward T1 measurement (vide infra). Solid state NMR is known to be a powerful technique for investigating the conformation and packing of molecules since chemical shift and peak line-broadening are both sensitive to local conformations. Additionally, some advanced solid state NMR techniques may give access to the spatial connectivity by using the dipolar interaction and the spin diffusion process.43 Adapted to 13C NMR, it was applied to better understand the structure and dynamics of surfactant molecules in mesophase silicates.44-46 The magnetization transfer measured on the 2,3C nuclei, for short contact times, indicates a rapid increase of the magnetization for tartrate anions. Since CP is more efficient for static 1H-13C dipolar interactions, this may suggest a more rigid carbon backbone for intercalated tartrate anions than for succinate anions. Furthermore, the conservation of the magnetization transfer (20% of the total magnetization for tCP = 20 ms) for long CP time with tartrate, while it is quenched for tCP greater than 10 ms with succinate, can be explained by a proton magnetization transfer relay acting over long distances in the case of tartrate. The magnetization transfer was also measured on the carboxylate group carbon 1,4C nuclei with nondirect C-H bonds (see the Supporting Information). For short CP times, there is no difference between the two intercalated anions. This similarity means that the 1,4C-H dipolar couplings in the

close vicinity of the 1,4C nuclei are similar as, therefore, are their environments. The C-H vectors involved must be 1,4C-CH within the molecule and 1,4C-O-HO-M resulting from the guest-host interaction both happening two bonds apart. On the other hand, for long CP times, 1,4C nuclei for tartrate anions sustain a greater transfer magnetization that may originate from intramolecular transfer obtained at longer distance, for instance, 1,4C-2,3CH-OH occurring three bonds apart. Yet, since 20% of the total magnetization is still observed for a CP time of 50 ms, we can surmise again that this transfer is likely to arise from a hydrogen-bonded subnetwork. The dynamic of interlayer water molecules was also investigated through the calculation of the diffusion coefficients from the Einstein relation (see the Supporting Information for completeness). Figure 5 shows the different components of the rootmean-square displacements as a function of time for Suc-Zn2Al. The data show that that the diffusion coefficient in the direction normal to the layer is smaller than those calculated in the plane parallel to the hydroxide layers for the two intercalated dianions. Furthermore, the diffusion coefficient of water calculated for SucZn2Al is slightly larger than that obtained with Tar-Zn2Al. This is consistent with the fact that water molecules are distributed more homogeneously over the interlayer region of Suc-Zn2Al and may be an additional contribution for higher protonic conductivity observed for this compound at room temperature as demonstrated hereafter (σ = 3.485  10-5 1/(Ω cm); see Table S3) . Ionic conduction in LDHs is generally explained as resulting from the proton transfer along the hydrogen bond network formed by the hydroxylated framework and the interlamellar water molecules through Grotthus and vehicle-type mechanisms. The intrinsic conductivity of the samples given in Figure 6 is reported in an Arrhenius diagram, displaying the evolution of log(σT) versus 1/T. For ClZn2Al and Suc-Zn2Al, a deviation from linearity is observed with a decrease in the slope as the temperature increases. Kreuer47,48 has investigated in detail the phenomenon of proton conductivity in watercontaining materials and clearly demonstrated that the non-Arrhenius behavior of transport properties, as observed for Cl-Zn2Al and SucZn2Al, is typical of disordered and weakly hydrogen bound networks generally associated with high proton conductivity. On the other hand, the Arrhenius law is obeyed rather satisfactorily from room temperature to -20 C in the case of Tar-Zn2Al with an activation energy of 0.44 eV. In a recent study, Frunza et al.49 pointed out that such low activation energies argue against a simple proton hopping mechanism 1487

dx.doi.org/10.1021/cm1030848 |Chem. Mater. 2011, 23, 1482–1490

Chemistry of Materials

Figure 6. Arrhenius diagrams of Cl-Zn2Al (filled circle), Tar-Zn2Al (filled square), and Suc-Zn2Al (open square). One cycle of temperatures is presented consisting of three steps: decrease of the temperature from room temperature to 253 K, then increase from 253 to 343 K, and finally, return to room temperature. (inset) Apparent activation energies derived from the slope of the curves log(σT) = f(1/T) in the temperature range 253-273 K. In Table S3 are reported the apparent activation energies (Ea), the conductivities (σ), and the charge carrier concentrations (σ0) determined at room temperature for the three samples.

and discussed a defect-related one, involving oxygen atoms not involved in hydrogen bonds; the jump of such defects occurs by a 120 rotation of water molecules. Thus, although the exact mechanism of conduction is not elucidated, the rotational dynamics of water molecules which depends strongly on the interlayer composition plays in every case a major role. Hence, Tar-Zn2Al must display a less disordered interlayer space with a strengthening of the hydrogen bonding that stabilizes the position of the interlayer species but hinders the diffusion and rotation/reorientation process thus resulting in a lower conductivity (σ = 1.766  10-5 1/(Ω cm); see Table S3). The relatively high crystallinity of Tar-Zn2Al led us to attempt a Rietveld refinement of the X-ray diffraction pattern and to test the structural model derived from the MD simulation. The refinement of the structure was carried out in the same space group as Cl-Zn2Al, i.e., R3m, and the cell parameter a was also that of the original cell since the tartrate anion can fit into this small cell. The input structure for the tartrate anions was obtained by averaging over conformations collected along the evolution of time of the MD simulation. Owing to the favorable interaction existing between the carboxylate and hydroxyl groups with the water molecules, the water molecules were included in the same positions. The graphical output of the refinement and atomic positions are reported in Figure 7 and Table S4, respectively. It must be noted that the application of the correction for anisotropic peak broadening using spherical harmonics was essential to reach a good fit, although the application of this model certainly blindly corrects for the intensity misfits in the pattern. The refinement of the atomic coordinates of the tartrate anion is revealed to be unstable with strong shifts of the atom positions. It is important to note that this averaged atomic coordinates obtained by MD are a static representation of the variability of the orientation of tartrate anions within the interlayer space. In these conditions, high displacement of the atomic positions are expected as a consequence of orientation heterogeneity and not due to uncertainty in the atomic

ARTICLE

Figure 7. Results of the structural refinement for Tar-Zn2Al: experimental X-ray diffraction (cross), calculated (line), Bragg reflections (ticks of the bottom are the ones of ZnO), and difference profiles.

Figure 8. Thermodynamic cycle used to calculate the relative enthalpy of exchange between the Tar-Zn2Al and Suc-Zn2Al LDHs. ΔrH(Cl 1 22f 1/2Suc ) = 2.8 kJ/mol, ΔrH2(Cl f 1/2Tar ) = -1.0 kJ/mol.

coordinates. This positional disorder is also visible on the atomic displacement parameters which were here constrained to a single overall parameter for the interlayer content and refined to 9.6(7) Å2. The water amount was refined to the realistic value of 4.5 water molecules per formula unit, although higher than the value given by the chemical analysis 3 and distributed mainly at the outer part of the interlayer (95%) in agreement with MD results. Energetics. We now focus on the accuracy of the potential used by the calculation of specific energy contributions. In the case of the Cl- f 1/2Suc2- exchange, the enthalpy change ΔrH1 determined by titration microcalorimetry is endothermic and equal to 2.8 kJ/mol whereas the process is exothermic for the Cl- f 1/2Tar2- exchange with ΔrH2 = -1.0 kJ/mol.50 Although the simulations methods are powerful technique to provide physical insights at the microscopic level, they require to be calibrated from an energy viewpoint. The difference in the enthalpies of exchange ΔΔrH was calculated from the thermodynamic cycle described in Figure 8. The relative enthalpy change can be expressed as ΔΔr H ¼ Δr H2 - Δr H1 ¼ ΔH1 - ΔH2

ð1Þ

where ΔΔrH = -3.8 kJ/mol. ΔH1 and ΔH2 are the enthalpy differences between tartrate and succinate in the solid state and in the aqueous phase, respectively (see Figure 8). They can be carried out either by finite differences of energy contributions between the different pairwise contributions in the tartrate and 1488

dx.doi.org/10.1021/cm1030848 |Chem. Mater. 2011, 23, 1482–1490

Chemistry of Materials succinate systems or by free energy perturbations51,52 and thermodynamic integration53,54 methods. The use of perturbation or integration thermodynamic methods cannot be applied efficiently in systems showing both a nonhomogeneous density distribution and a high degree of confinement due to the lack of convergence of the calculated free energy. We take the route of calculating finite differences between the different contributions listed in Table S7 for the Lennard-Jones and Coulomb interactions. These energy contributions are calculated in Suc-Zn2Al and Tar-Zn2Al for an ordering and disordering distribution of the trivalent cations within the hydroxide layers and are reported in Table S7. Table S8 reports the differences of the energy contributions calculated between Tar-Zn2Al and Suc-Zn2Al. Summing then these energy differences leads to ΔH2 = -0.07 kJ/mol. This weak value means that the hydration of the two anions is sensitively the same: the number of water molecules in the first shell of hydration is equal to 24.3 and 25.1 for the tartrate and succinate anions, respectively. We observe from the ΔH1 values of Table S8 that the difference in the enthalpy of exchanges between the Tar-Zn2Al and Suc-Zn2Al is accurately reproduced within the fluctuations inherent to this type of calculation. The value of the predicted enthalpy difference is in the energy range [-2.8,-0.1] kJ/mol depending on the way of distributing the metal cations. This straightforward calculation makes the force field consistent from an energy viewpoint. We also note that the major energy contribution comes from the dianion-dianion pairwise contribution indicating a specific interaction between adjacent tartrate anions that does not exist between adjacent succinate anions. This is in line with previous conclusions drawn from microcalorimetric enthalpy values50 and explains the difference in the rotational dynamics between the tartrate and succinate anions.

’ CONCLUSIONS The energy and structural characterization of the interlayer domain is a key parameter for determining the structure of hybrid-LDHs. The properties of the interlayer can be defined in terms of interlayer distance, dynamics of the intercalated species, and cation ordering within the layer. They depend on the balance between the different energy contributions (van der Waals, electrostatic, hydrogen bond) within the LDH. Although the molecular modeling is a powerful technique, it remains then essential to validate the force field and methodology used with complementary experimental data. The resulting combined approach is then very challenging and allows a better understanding of the structureproperty relationship in an LDH system. We have followed this strategy to study the property of a Zn2Al LDH intercalated with tartrate and succinate anions. These organic species can be viewed as model molecules to investigate in particular the impact of the hydroxyl groups on the organization and the properties of the interlayer domain. The comparison between the simulated and electronic density profiles is excellent. A more hydrophilic central zone in the interlayer domain is established for intercalated tartrate molecules. The MD simulations, NMR, and proton conductivity are in line to establish a rotation of the succinate anions around the C1-C4 axis and no rotation for the tartrate anions. As a consequence, the diffusion coefficient of water molecules as well as the protonic conductivity is slightly larger for SucZn2Al than for Tar-Zn2Al. We have completed the characterization of the interlayer domain by the calculation of the difference in the enthalpies of exchange between Tar-Zn2Al and Suc-Zn2Al. The

ARTICLE

sign of the enthalpy difference is consistent with the microcalorimetric enthalpy values and establishes the accuracy of the force field used. The energy calculations show that the hydration of the two anions is sensitively the same. They also evidence the existence of more favorable tartrate-tartrate interactions that explain the difference between the tartrate and succinate dynamics. The integrated theoretical and experimental approach proposed here allow us to demonstrate for the first time the importance of the hydrogen-bond network inducing a specific orientation of tartrate anions with respect to the hydroxide layers and surrounding molecules with major consequences on the dynamics of the interlayer species and overall LDH material reactivity. We believe that such combined approach and the results obtained here can allow a better understanding of the nanostructure of other hybrid LDH intercalated with more complex anions.

’ ASSOCIATED CONTENT

bS

Supporting Information. Descriptions of experimental techniques (chemical analyses of the phases, samples characterization by X-ray diffraction, NMR and complex impedance spectroscopies, Rietveld refinement) and computational methods (potential models, structure, orientation, and dynamics). This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected] (C.T.-G.); [email protected] (P.M.).

’ ACKNOWLEDGMENT One of the authors, P.M., is indebted to F. Goujon, A. Ghoufi, and Y. Israeli for helpful discussions. This work was granted access to the HPC resources of IDRIS under the allocation 2010i2010092119 made by GENCI (Grand Equipement National de Calcul Intensif). ’ REFERENCES (1) Romero, P. G.; Sanchez, C. Functional Hybrid Materials; Wiley: New-York, 2004. (2) Latterini, L.; Nocchetti; Aloisi, G. G.; Costantino, U.; Elisei, F. Inorg. Chim. Acta 2007, 360, 128. (3) Del Hoyo, C. Apply. Clay. Sci 2007, 36, 103. (4) Lang, K.; Bezdieka, P.; Bourdelande, J. L.; Hernando, J.; Jirka, I.; Kafunkova, E.; Kovanda, F.; Mosinger, J.; Wagnerova, D. Chem. Mater. 2007, 19, 3822. (5) Desigaux, L.; Belkacem, M. B.; Richard, P.; Cellier, J.; Leone, P.; Cario, L.; Leroux, F.; Taviot-Gueho, C.; Pitard, B. Nano Lett. 2006, 6, 1999. (6) Mousty, C. Appl. Clay. Sci. 2004, 27, 159. (7) Choi, S. J.; Oh, J. M.; Choy, J. H. J. Nanosci. Nanotechnol. 2008, 8, 5297. (8) Leroux, F. J. Nanosci. Nanotechnol. 2006, 3, 303. (9) Leroux, F.; Besse, J.-P. Chem. Mater. 2001, 13, 3507. (10) Ding, P.; Chen, W.; Qu, B. Prog. Nat. Sci. 2006, 16, 573. (11) Aicken, M.; Bell, I. S.; Coveney, P. V.; Jones, W. J. Adv. Mater. 1997, 9, 496. (12) Newman, S. P.; Williams, S. J.; Coveney, P. V.; Jones, W. J. Phys. Chem. B 1998, 102, 6710. (13) Fogg, A. M.; Rohl, A. L.; Parkinson, G. M.; O’Hare, D. Chem. Mater. 1999, 11, 1194. 1489

dx.doi.org/10.1021/cm1030848 |Chem. Mater. 2011, 23, 1482–1490

Chemistry of Materials (14) Wang, J.; Kalinichev, A. G.; Kirkpatrick, R. J.; Hou, X. Chem. Mater. 2001, 13, 145. (15) Trave, A.; Selloni, A.; Goursot, A.; Tichit, D.; Weber, J. J. Phys. Chem. B 2002, 106, 12291. (16) Newman, S. P.; Coveney, T. D. C. P. V. Langmuir 2002, 18, 2933. (17) Greenwell, H. C.; Jones, W.; Newman, S. P.; Coveney, P. V. J. Mol. Struct. 2003, 647, 75. (18) Lombardo, G. M.; Pappalardo, G.; Punzo, F.; Costantino, F.; Costantino, U.; Sisani, M. Eur. J. Inorg. Chem. 2005, 5026. (19) Mohanambe, L.; Vasudevan, S. Langmuir 2005, 21, 10735. (20) Mohanambe, L.; Vasudevan, S. J. Phys. Chem. B 2005, 109, 15651. (21) Kumar, P. P.; Kalinichev, A. G.; Kirkpatrick, R. J. J. Phys. Chem. B 2006, 110, 3841. (22) Kumar, P. P.; Kalinichev, A. G.; Kirkpatrick, R. J. J. Phys. Chem. C 2007, 111, 13517. (23) Thyveetil, M. A.; Coveney, P. V.; Suter, J. L.; Greenwell, H. C. Chem. Mater. 2007, 19, 5510. (24) Thyveetil, M. A.; Coveney, P. V.; Greenwell, H. C.; Suter, J. L. J. Am. Chem. Soc. 2008, 130, 4742. (25) Pisson, J.; Morel, J. P.; Morel-Desrosiers, N.; Taviot-Gueho, C.; Malfreyt, P. J. Phys. Chem. B 2008, 112, 7856. (26) Zhang, H.; Xu, Z. P.; Lu, G. Q.; Smith, S. C. J. Phys. Chem. C 2009, 113, 559. (27) Pisson, J.; Taviot-Gueho, C.; Isra€eli, Y.; Leroux, F.; P. Munsch, P.; Itie, J. P.; Briois, V.; Morel-Desrosiers, N.; Besse, J. P. J. Phys. Chem. B 2003, 107, 9243. (28) Prevot, V.; Forano, C.; Besse, J.; Abraham, F. Inorg. Chem. 1998, 37, 4293. (29) Whittingham, M.; Jacobson, A. Intercalation Chemistry; Academic press: New-York, 1982. (30) Mayo, S. L. .; Olafson, D.; W., A. G., III J. Phys. Chem. B 1990, 94, 8897. (31) Rappe, A. K.; Casewit, C. J.; Colwell, K. S.; W., A. G., III; Skiff, W. M. J. Am. Chem. Soc. 1992, 114, 10024. (32) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D. J. Chem. Phys. 1983, 79, 926. (33) Breneman, C. P.; Wiberg, K. B. J. Comput. Chem. 1990, 11, 361. (34) Schmit, M. W.; Baldridge, K. B.; Boatz, J. A.; Elbert, S. T.; Gordon, M. S.; Jensen, J. J.; Koseki, S.; Matsunaga, N.; Nguyen, K. A.; Su, S.; Windus, T. L.; Dupuis, M.; Montgomery, J. A. J. Comput. Chem. 1993, 65, 14. (35) Allen, M. P.; Tildesley, D. J. Computer Simulation of Liquids; Clarendon Press: Oxford, 1989. (36) Smith, E. R. Proc. R. Soc. Lond. A. 1981, 375, 475. (37) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, A.; DiNola, A.; Haak, J. R. J. Chem. Phys. 1984, 81, 3684. (38) DL_POLY is a parallel molecular dynamics simulation package developed at the Daresbury Laboratory Project for Computer Simulation under the auspices of the EPSRC for the Collaborative Computational Project for Computer Simulation of Condensed phases (CCP5) and the Advanced Research Computing Group (ARCG) at the Daresbury Laboratory. (39) Isra€eli, Y.; Taviot-Gueho, C.; Besse, J. P.; Morel, J. P.; MorelDesrosiers, N. J. Chem. Soc., Dalton Trans. 2000, 791. (40) Radoslovich, E. W. Am. Mineral. 1963, 48, 76. (41) Prevot, V. Interlayer LDH Domain Study and Grafting Reaction. Ph.D. Thesis, Blaise Pascal University, 1999. (42) Kolodziejski, W.; Klinowski, J. Chem. Rev. 2002, 102, 613. (43) Bonhomme, C.; Coelho, C.; Baccile, N.; Gervais, C.; Azais, T.; Babonneau, F. Acc. Chem. Res. 2007, 40, 738. (44) Wang, L. Q.; Liu, J.; Exarhos, G. J.; Bunker, B. C. Langmuir 1996, 12, 2663. (45) Simonutti, R.; Comotti, A.; Bracco, S.; Sozzani, P. Chem. Mater. 2001, 13, 771. (46) Wang, L.-Q; E., G. J.; Liu, J.; Flanigan, K. Y.; Bordia, R. J. Phys. Chem. B 2000, 104, 2810.

ARTICLE

(47) Kreuer, K. D. Chem. Mater. 1996, 8, 610. (48) Kreuer, K. D. Solid State Ionics 2000, 136, 1469. (49) Frunza, L.; Sch€onhals, A.; Frunza, S.; Parvulescu, V. I.; Cojocaru, B.; Carriazo, D.; Martin, C.; Rives, V. J. Phys. Chem. A 2007, 111, 5166. (50) Morel-Desrosiers, N.; Pisson, J.; Isra€eli, Y.; Taviot-Gueho, C.; Besse, J. P.; Morel, J. P. J. Mater. Chem. 2003, 13, 2582. (51) Zwanzig, R. W. J. Chem. Phys. 1954, 22, 1420. (52) Mezei, M.; Beveridge, D. L. Ann. N.Y. Acad. Sci. USA 1986, 482, 1. (53) Mezei, M.; Swaminathan, S.; Beveridge, D. L. J. Am. Chem. Soc. 1978, 100, 3255. (54) Pearlman, D. A. J. Comput. Chem. 1994, 15, 105.

1490

dx.doi.org/10.1021/cm1030848 |Chem. Mater. 2011, 23, 1482–1490