QENS and NMR Study of Water Dynamics in SBA-15 with a Low Water

Article pubs.acs.org/JPCC

QENS and NMR Study of Water Dynamics in SBA-15 with a Low Water Content Anna Kiwilsza,†,‡ Aleksandra Pajzderska,† Miguel A. Gonzalez,§ Jadwiga Mielcarek,∥ and Jan Wąsicki*,†,‡ †

Faculty of Physics and ‡NanoBioMedical Center, Adam Mickiewicz University, Umultowska 85, 61-614 Poznań, Poland § Institute Laue-Langevin, B.P. 156x, 38042 Grenoble Cedex 9, France ∥ Department of Inorganic and Analytical Chemistry, Poznan University of Medical Sciences, Grunwaldzka 6, 60-780 Poznan, Poland S Supporting Information *

ABSTRACT: In this study, motions performed by water molecules adsorbed on the silica surface of SBA-15 material with 6.1% of water content (15% of pore filling) were investigated using NMR and Quasielastic Neutron Scattering (QENS) techniques. The results show no sign of translational diffusion of water, but two types of stochastic localized motions were identified, and both described using a model of proton jumps between two sites. For both motions, the characteristic jump distances and correlation times, as well as activation energies, have been extracted and found to differ significantly. On this basis, the faster motion was ascribed to jumps of water molecules between neighboring positions (d = 2.5 Å, τ = 4 ps at 300 K, and Ea = 5.2 ± 0.2 kJ/mol from NMR data, and 5.6 ± 1.1 kJ/mol from QENS), while the slower one exhibits a temperature dependent jump distance and was ascribed to jumps of water molecules between more spatially separated positions (d = 2.9−4.3 Å, τ = 25 ps at 300 K, and Ea = 16.1 ± 0.3 kJ/mol from NMR, and 17.3 ± 0.3 kJ/mol from QENS data).

I. INTRODUCTION The properties of confined water in silica pores has been studied extensively for many years using different experimental techniques such as X-ray diffraction,1,2 differential scanning calorimetric,3,4 nuclear magnetic resonance (NMR),5−8 and neutron scattering.2,9−11 Also, molecular dynamics simulations have been widely used.12,13 These investigations have shown that confined water has different structural and dynamical properties from those of bulk water. It should be underlined that most studies have mainly focused on overfilled and fully filled pores. In general, confined water can perform two types of motion: translational and rotational diffusion. NMR as well as quasi-elastic neutron scattering (QENS) measurements have shown that the translational diffusion of water in the confined space of pores is slower than that of bulk water.5,9,10 The degree of diffusion slowdown depends on the size of the pores and the hydration level. However, not much attention has been paid to characterizing the behavior of small amounts of water absorbed in silica materials, even if from the point of view of different porous silica applications (especially as drug delivery systems), an important issue is the presence of water molecules at small water content inside their pores. Some attempts at describing the structure of water molecules at a silica surface have been made by Buntkowsky et al. by NMR methods.14 Other authors15 presented molecular dynamics simulations of confined water in a silica pore in the low hydration level, revealing strong layering effects and a strong distortion of the hydrogen bond network. © 2015 American Chemical Society

The small water content can influence the behavior of other molecules confined in silica pores and therefore have repercussions in such applications. We consider therefore that is important to determine how water behaves at small concentrations (e.g., is water free to diffuse or not) as a first step, before analyzing in a second step the interplay between the matrix, the water molecules present in SBA and the host molecules. The aim of this study was to perform a thorough analysis of water dynamics confined in mesoporous silica SBA15 at a small water content, using NMR and QENS methods. SBA-15 is characterized by an ordered cylindrical porous structure, good sorption properties, and thermal stability. It has mesopores of diameters ranging from 5 to 10 nm and micropores formed during the process of synthesis.16,17 As the surface of SBA-15 is hydrophilic, it adsorbs easily water molecules from air.18 As mentioned above, no detailed description of water dynamics at a small water content has been made up to now, so our study focused on the characterization of the water dynamics including information about the time scale and the geometry of their motions. In section II of this paper we describe briefly our experimental methods, in section III we present the main results of our study together with their discussion, and section IV contains conclusions. Received: March 19, 2015 Revised: June 27, 2015 Published: June 29, 2015 16578

DOI: 10.1021/acs.jpcc.5b02672 J. Phys. Chem. C 2015, 119, 16578−16586

Article

The Journal of Physical Chemistry C

II. EXPERIMENTAL METHODS II.1. SBA-15 Preparation. Mesoporous silica material SBA15 was synthesized by the method proposed by Zhao et al.16 T h e t e m p l a t e w a s a n o n i o n i c s u r fa c t a n t ( H O (CH2CH2O)20(CH2CH(CH3)O)70(CH2CH2O)20H) known under the commercial name Pluronic P123 (BASF Corporation), while the silica source was tetraethoxysilane TEOS (Remmers Baustofftechnik GmbH). The synthesis was carried out in a water environment of acidic pH at 308 K for 20 h. The mixture was postsynthetically aged at 378 K for 24 h, then filtered off and twice washed with distilled water. The final product was calcined in air at 773 K for 6 h and dried at 323 K for 48 h. The water was adsorbed by natural process, and its amount was controlled just before performing the experiments by thermogravimetry analysis. II.2. Basic Characterization of SBA-15. Sorption properties of the product were tested by low-temperature nitrogen sorption (analyzer ASAP 2010, Micrometrics). The nitrogen adsorption isotherm was classified as type IV, according to IUPAC, and gives a BET specific surface area of 710.25 m2/g, a BJH total pore volume of 0.44 cm3/g and a mean pore size of 5.55 nm. The hexagonal pore arrangement typical of SBA-15 was confirmed by small-angle X-ray diffraction (AXS D8 Advance diffractometer, λ(Cu) = 1.5406 Å, 35 kV, 50 mA, Bruker). From the SAXS diffractogram we determined the lattice constant as a = 9.67 nm and the mean thickness of the walls of the silica matrix as t = 4.12 nm. These values are in agreement with those obtained from transmission electron microscopy images (JEM-1400, Jeol microscope). The amorphous character of the silica building the walls of the mesopores was confirmed by wide-angle X-ray diffraction (Empyrean diffractometer, λ(Cu) = 1.5406 Å, 40 kV, 30 mA, PANalytical). Differential scanning calorimetry (DSC) measurements were made in the range from room temperature to 673 K using a differential calorimeter DSC-50 (Shimadzu). A sample of 3.05 mg was mounted in an aluminum nonhermetically closed pan of 40 μL in capacity and heated at a rate of 10 K/min in a nitrogen atmosphere at the flow rate of 30 mL/min. Temperature and heat flow were measured with an accuracy of 0.01 K and 0.001 mW, respectively. Fourier transform infrared spectroscopy (FTIR) spectra of solid samples at room temperature were recorded on a spectrometer FT-IR IFS 66v/S (Bruker) by the transmission method in the wavenumber range 4000−400 cm−1 and with a resolution of 1 cm−1. The sample containing a homogeneous mixture of 1.5 mg SBA15 with 200 mg KBr was made into tablets of 13 mm in diameter. Thermogravimetric analysis (TGA) was made at the Laboratory for Structural Studies at the Faculty of Physics, AMU on a thermogravimeter TGA Q50 V20.7 Build 32 (TA Instruments). A sample of 10 mg was placed on a platinum plate and heated from room temperature to 1173 K at the rate of 10 K/min in a nitrogen atmosphere. Temperature and mass were measured with accuracies of 10−5 K and 10−5 mg, respectively. II.3. Nuclear Magnetic Resonance. 1H NMR spectra were recorded on a pulse spectrometer operating at a resonant frequency of 58.9 MHz constructed at the Radiospectroscopy Division at the Faculty of Physics, AMU. The temperature of the sample was controlled by means of a gas-flow cryostat and monitored with a Pt resistor with an accuracy of 1 K. The SBA15 sample for 1H NMR studies was sealed off in a glass ampule. After application of a π/2 pulse of 3.6 μs, the free induction

decay (FID) signal was recorded. The intensity of the FID signal was determined in 30 μs after the pulse. As the dead time of the measuring head was 10 μs, and the FID signal from ice disappeared after a time shorter than 20 μs, only the signal from liquid water was recorded. From the width of the FID signal at its half-maximum the apparent spin−spin relaxation time T2* was determined. Measurements were performed in the range from room temperature to 110 K. II.4. Quasielastic Neutron Scattering. Quasielastic neutron scattering (QENS) measurements were performed using two different instruments: the backscattering spectrometer IN16 (λ = 6.27 Å, energy resolution fwhm = 0.9 μeV, Q range 0.29−1.92 Å−1, energy transfer range ±14.5 μeV) and the time-of-flight spectrometer IN5 (λ = 6 Å, energy resolution fwhm = 35 μeV, Q range 0.2−2.0 Å−1), both at the LaueLangevin Institute in Grenoble (France). Because of the low density of the material studied (approximately 0.3 g/cm3), a sample of 1.03 g SBA-15 was placed in a flat aluminum container of size 30 mm × 40 mm and 2.0 mm in thickness, to ensure a transmission coefficient of 0.9. An elastic fixed window scan was recorded on IN16 scanning on temperature between 250 and 10 K at a mean cooling rate of 2 K/min. Quasielastic neutron scattering spectra were then measured at 100, 150, 210, 250, 300 (IN5), and 210 K (IN16). To obtain the background and the resolution of the two spectrometers, the spectrum was recorded for an empty cell and for a 1 mm thick vanadium foil, respectively. The angle between the sample and the incident neutron beam was set to 135°. The sample temperature was controlled by a helium−nitrogen cryostat with an accuracy of 0.01 K. The experimental data were corrected (subtraction of background from empty cell, correction for detector efficiency, normalization to the vanadium spectrum and correction for absorption) using LAMP.19

III. RESULTS AND DISCUSSION III.1. Preliminary Study: DSC, TGA, and FTIR. The presence of water molecules adsorbed on silica surface was confirmed using DSC, TGA, and FTIR methods (see Supporting Information, Figures 1S, 2S, and 3S, respectively). A broad endothermic peak on the DSC curve with a maximum at 313.8 K corresponds to the gradual process of release of water adsorbed on the surface of the silica matrix.18 It characterized by a heat of transition less than 50 J/g, which suggested a low level of water content in the studied material. A detailed interpretation of the infrared spectrum of SBA-15 has been given in ref 20. The recorded spectrum reveals exactly the same features, typical for mesoporous silica, with a characteristic silicon−oxygen bond vibrations covering the range ∼1250−1000 cm−1. In addition, the symmetric stretching vibrations and the bending modes of the Si−O−Si bridges contribute at about 810 and 450 cm−1. The recorded spectrum reveals very prominent bands assigned to water vibrations, which were found at 563, 1643, and 1886 cm−1. The quoted modes correspond to water librations (L), internal bending (δ), the association (A) modes (anharmonic combination of bending, libration, and hindered translation: A = δ + L − T).21 Furthermore, the characteristic broad band appears around ∼3500 cm−1, being dominated by the OH symmetric and antisymmetric water stretching contributions. If referring to the spectra of water in liquid and solid-state, one can note that the band is blue-shifted and considerably broadened toward higher 16579


Article


240 K) and low value (below 120 K) are separated by an intermediate region (240−120 K), where a quite sharp decrease of intensity is observed. This decrease in our sample can be interpreted as the result of a gradual freezing of water inside the pores and a reduction of the mobility of the water molecules, which could be due either to a general slowdown of the water dynamics or to a reduction in the amount of water molecules that retain some mobility in the investigated time scale.24 T2* Relaxation. In the temperature region where the FID signal intensity decreases (240−120 K), the apparent spin−spin relaxation time T2* was determined experimentally. Figure 2

wavenumbers. It may suggest that the water associates are more loosely bound. The upper side of the band (>3500 cm−1) can be attributed to the contributions of the hydrogen-bonded water molecules adsorbed on the silica surface.20−22 Finally, one can note a barely visible, upper-edge shoulder around ∼3750 cm−1, which is due to the ν(OH) vibrations of the isolated hydroxyl moieties. The content of water and surface silanol groups in SBA-15 was evaluated quantitatively by thermogravimetry (Figure 3S). A characteristic sharp decrease corresponding to a mass loss of 6.1% is observed when heating from room temperature to 423 K and interpreted as a result of dehydration. The continuous mass loss up to 7.8%, taking place from 423 to 1173 K, is related to a gradual decomposition of silanol groups SiOH, by their condensation in vapor (dehydroxylation).18,23 The mass loss related to dehydration and dehydroxylation indicates that the number of water molecules and OH groups per 1 g SBA-15 are 22.1 × 1020 H2O/g and 12.3 × 1020 OH/g, respectively. Therefore, on average there are 1.8 water molecules adsorbed on the silica surface per OH group in the system and the number of OH groups per unit surface area is 2.0/nm2, which is consistent with literature data.23 The degree of filling of the SBA-15 pores with water molecules calculated on the basis of the nitrogen sorption results is 15% of the pore volume. III.2. NMR Measurements. FID Signal Intensity. When cooling from room temperature, initially the “as-measured” intensity of the FID signal increases slightly, then, between 240 and 120 K, we observe a significant decrease of intensity, and finally, below 120 K, the intensity reaches a small, almost constant value. The general increase in the FID signal intensity with decreasing temperature (in the whole temperature range) is described by the Curie law which states that the intensity is inversely proportional to temperature. This law concerns all the protons in the sample and does not depends on the type of material studied. In order to exclude from the results the effect of the increase in the FID signal intensity with decreasing temperature (in consistence with the Curie law), at each temperature the “asmeasured” FID intensities were multiplied by T/273. Figure 1 presents the temperature dependence of the FID signal intensity after this correction related to the Curie law. Two regions with almost constant intensities of high value (above

Figure 2. Apparent spin−spin relaxation time T2* vs reciprocal temperature for SBA-15.

presents the plot of log T2* versus reciprocal temperature, where two regions are observed. The slopes of the linear sections of log T2* as a function of 1/T bring information on the activation energies of these processes, which are 16.1 ± 0.3 kJ/mol (between 240 and 170 K) and 5.2 ± 0.2 kJ/mol (between 160 and 120 K), respectively. In order to establish the type of proton dynamics from water molecules inside the SBA-15 pores, we performed QENS measurements. III.3. QENS Study. Elastic Fixed Window Scan. The temperature dependence of the elastic intensity measured on IN16 between 10 and 250 K is shown in Figure 3. The system shows a harmonic behavior up to about 175 K, where an additional decrease of the elastic intensity related to the activation of some motions (due to the onset of a quasielastic component) is observed. The decrease in the intensity of the elastic component of the spectrum is described by the equation: ⎛ ⟨u(T )2 ⟩Q 2 ⎞ I(Q⃗ , T ) = I(0)exp⎜ − ⎟ 3 ⎝ ⎠

(1)

where I(0) is the intensity at zero temperature, while exp(−((⟨u(T)2⟩Q2)/3)) is the Debye−Waller factor depending on the scattering vector Q⃗ . The temperature dependence of the mean square displacements ⟨u(T2)⟩ was obtained from the slope of −ln I(Q) versus Q2 in the range 0.19−1.92 Å−1 for each temperature. As follows from the plot ⟨u(T2)⟩(T) in Figure 3, with increasing temperature the amplitude of thermal vibrations of protons increases. Moreover, the appearance of a quasielastic component producing a sharper decay of the elastic intensity above 175 K leads to a larger increase in the mean

Figure 1. Temperature dependence of the FID signal intensity for SBA-15. 16580


Article


be fitted by one single Lorentzian function (n = 1). However, as both spectrometers have very different energy resolutions, this implies that there are two different types of motion in the system taking place in different time scales. For temperatures above 210 K, the fitting of the IN5 data required two Lorentzian functions (n = 2), implying that the times characterizing both types of motion were close enough to be detectable on the same spectrometer. The EISF and QISF values determined from these fits at 210 (n = 1), 250, and 300 K (n = 2) are shown in Figure 5. Additionally, the spectra for temperatures 250 and 300 K were reanalyzed including three Lorentzian functions (n = 3). Although for some Q values it is possible to add a third component, such fits did not provide a consistent picture and therefore were rejected. We observe that with increasing temperature, the EISF decreases. Moreover, the minimum observed in the EISF curve is shifted toward smaller values of the scattering vector Q⃗ . Thus, at 300 K the minimum is clearly visible at Q = 1.2 Å−1, while at lower temperatures, the EISF curves get flattened in the range from 0.4 to 1.8 Å−1, making difficult to extract the exact position of the minimum on the basis of experimental data alone. The changes observed in the EISF imply that with increasing temperature, the number of protons taking part in a motion increases, while the characteristic radius of the motion changes. Additional information can be obtained from the Q dependence of the QISF. At all temperatures, the QISF branch related to the Lorentzian function of smaller width (Figure 5, open circles) shows a maximum. At 300 K, the maximum appears at Q = 1.0 Å−1, while for lower temperatures it is again difficult to determine accurately the position of the maximum. However, it can be concluded that with increasing temperature the maximum is shifted toward smaller values of the scattering vector Q⃗ , implying that the characteristic radius of motion described by the Lorentzian function of smaller width increases with temperature. The temperature dependence of the QISF branch corresponding to the Lorentzian function of greater width (Figure 5, open triangles) does not reveal any maximum for all temperatures in the Q-range covered by IN5 (0.4−1.8 Å−1). This observation implies that the characteristic radius of this motion is small in comparison with that of the motion described by the Lorentzian function of smaller width. Finally, the analysis of the half-widths Γ of the Lorentzian functions describing the quasielastic broadening shows that there is not any significant change of this parameter as a function of the scattering vector (Figure 4S). The average halfwidths obtained for the two types of motion observed at each temperature are 98.6 and 1.5 μeV (210 K, IN5 and IN16, respectively), 149.6 and 15.1 μeV (250 K, IN5), and 350.8 and 53.8 μeV (300 K, IN5). Models of Motion. The behavior of the EISF and QISF and the fact that the widths of the fitted Lorentzian functions do not show any marked dependence with Q indicate that water molecules are not able to perform long-range diffusion in our system and that we are observing two localized motions. In order to determine the possible nature of such motions we consider two different models that are compatible with the information obtained from the previous analysis, that is, with the Q-dependence of the EISF, QISFs, and quasielastic widths. The first model corresponds to a jump between two equivalent positions (2SJ). The theoretical scattering function for this model takes the following form:25

Figure 3. Temperature dependence of the elastic intensity summed over all the detectors in IN16, covering a range 0.19 to 1.92 Å−1 of the neutron scattering spectrum (squares) and mean squared displacement of thermal vibrations of protons (triangles). Lines are guides for the eye.

square displacement, although the maximum value of the amplitude does not exceed 0.7 Å2 at 240 K. QENS Measurements and Analysis. The dynamics of water in SBA-15 was explored by means of QENS spectra taken at five temperatures: 100 (IN5), 150 (IN5), 210 (IN5 and IN16), 250 (IN5), and 300 K (IN5). In this system the main contribution to scattering comes from hydrogen nuclei, so the observed spectrum S(Q⃗ ,ω) is proportional to the probability at which neutrons are scattered by the protons present in the SBA-15 sample (belonging either to surface silanol groups or to adsorbed water molecules). Initially, the experimental QENS spectra were fitted using the PeakFit program by means of the following general function: ⎛ ⟨u(T )2 ⟩Q 2 ⎞ S(Q⃗ , ω) = exp⎜ − ⎟ 3 ⎝ ⎠ ⎛ × ⎜⎜[A 0(Q⃗ )δ(ω) + ⎝

n

⎞

∑ Ai(Q⃗ )Li(ωi , Q⃗ , Γi)] ⊗ R(Q⃗ , ω)⎟⎟ i=1

+ B(Q⃗ , ω)

⎠ (2)

where exp(−((⟨u(T) ⟩Q )/3)) is the Debye−Waller factor, A0(Q⃗ ) is the elastic incoherent structure factor (EISF), δ(ω) is the Dirac delta function, Ai(Q⃗ ) is the quasielastic incoherent structure factor (QISF), Li(ωi,Q⃗ ,Γi) is the Lorentzian function describing the quasielastic line broadening, Γi is the half-width of Lorentzian function R(Q⃗ ,ω) is the resolution function of the spectrometer (described by a Gauss function for IN5 and IN16), and B(Q⃗ ,ω) is the background. In order to achieve the best fit, the procedure was performed iteratively using a different number of Lorentzian functions (from n = 0 to n = 2) to describe the quasielastic line broadening, as discussed below. As an example of the data obtained and the fit quality, Figure 4 presents selected experimental spectra and their fit to eq 2 for different temperatures and scattering vectors. At 100 K, no quasielastic broadening was observed (n = 0). But at 150 K, a very tiny quasielastic component could already be noticed (n = 1), suggesting that even below 175 K there is some mobility in the system, even if only a small part of the protons is involved in such motion. At 210 K, in the spectra recorded on IN16 and IN5, the quasielastic component could 2

2

16581


Article


Figure 4. Experimental neutron scattering spectra for SBA-15 and fits to the general scattering function given by eq 2 for (a) IN5, Q = 1.5 Å−1 at five different temperatures, and (b) IN16, T = 210 K at selected scattering wavevectors (black squares, experimental points; red solid line, fit model; black dot line, resolution curve; black solid line, background; green and blue solid lines, quasielastic components).

S2SJ(Q , ω) = A 02SJ(Q , d)δ(ω) + A12SJ(Q , d)L(ω , Γ)

The second model considered is that of isotropic diffusion in a restricted volume, called restricted diffusion (RD). The theoretical scattering function for this model takes the form:25,26

(3)

for which the EISF and QISF take the form: 1 A 02SJ(Q , d) = (1 + j0 (Qd)) 2

S RD(Q , ω) = A 0RD(Q , r )δ(ω) + A1RD(Q , r )L(ω , Γ)

(4)

(6)

1 (1 − j0 (Qd)) (5) 2 where j0(Qd) is the spherical Bessel function of zeroth order, while d is the jump distance. For this model the width of the Lorentzian function describing the quasielastic broadening does not depend on the scattering vector Q⃗ .25 A12SJ(Q , d) =

and its EISF and QISF are described by ⎛ 3j (Qr ) ⎞2 ⎟ A 0RD(Q , r ) = ⎜ 1 ⎝ Qr ⎠ 16582

(7) DOI: 10.1021/acs.jpcc.5b02672 J. Phys. Chem. C 2015, 119, 16578−16586

Article


where j1(Qr) is the first order spherical Bessel function and r is the radius of diffusion. In this model the width of the Lorentzian function describing the quasielastic broadening does not depend on the scattering vector Q⃗ for Q⃗ ≤ (π/r).26 As in our case we found that the width of the Lorentzian function describing the quasielastic broadening does not depend on Q at least up to 1.8 Å−1, the model of restricted diffusion could be used only if the diffusion radius obtained from the fit satisfies the condition: r ≤ 1.74A

(9)

Similar models have already been considered in the literature for systems with surface bound water, such as silica glasses of Vycor type,27 DNA,28 proteins,29,30 graphite oxide layers,31 or cyclodextrin.32 At 210 K, the spectra recorded on both IN5 and IN16 can be fitted with a single Lorentzian function, so they were fitted using either the model of jumps between two positions 2SJ or the model of restricted diffusion RD. Because the diffusion radii obtained from the fit based on the RD model (for IN5, r = 2.1 Å; for IN16, r = 2.8 Å) did not meet condition 9, this model was rejected from further analysis, and the 2SJ model was assumed as the best description of the protons motion at this temperature. The theoretical curves obtained from the fit of the EISF(Q) and QISF(Q) at 210 K using the 2SJ model (eqs 4 and 5, respectively) are shown in Figure 5a. As we also know that at 250 and 300 K the quasielastic component cannot be described by a single Lorentzian function, we analyzed a model made of the sum of the two above-mentioned models, assuming that individual motions take place in two independent subsystems. In this case, the contribution of the i-th motion in the total scattering is directly related to the number of protons performing such kind of motion (Ni). It was also assumed that a certain number of protons (denoted as N0) do not take part in any motion, so their presence enhances the amplitude of the elastic component of the spectrum. Three possible models composed of two types of motion were considered: (i) a model consisting of two types of jumps between two positions 2SJ + 2SJ, (ii) a model of jumps between two positions and restricted diffusion 2SJ + RD, and (iii) a model of two types of restricted diffusion having different characteristic radii RD + RD. However, the last two models (2SJ + RD and RD + RD) produce unphysical fitting parameters (r < 0.01 Å for both temperatures), so they were rejected. Therefore, only the first one (2SJ + 2SJ) was considered for further analysis. The theoretical scattering function obtained for this model is S(Q , ω) = A 0(Q , d1, d 2)δ(ω) + A1(Q , d1)L(ω , Γ1)

(10)

+ A1(Q , d 2)L(ω , Γ2)

Figure 5. EISF (closed symbols) and QISF (open symbols) obtained from the fit with the general scattering function given by eq 2 to the experimental spectra recorded at (a) 210, (b) 250, and (c) 300 K. At each temperature, two QISF branches were obtained described by the Lorentzian functions of smaller (circles) and greater (triangles) width. The theoretical curves correspond to a selected model of motion given by eqs 4 and 5 (210 K) and eqs 11−13 (250 and 300 K; description in the text).

⎛ 3j (Qr ) ⎞2 RD ⎟ A1 (Q , r ) = 1 − ⎜ 1 ⎝ Qr ⎠

where EISF and QISF are expressed as A 0(Q , d1, d 2) = N0 + N1A 02SJ(Q , d1) + (1 − N0 − N1)A 02SJ(Q , d 2)

(11)

A1(Q , d1) = N1A12SJ(Q , d1)

(12)

A1(Q , d 2) = (1 − N0 − N1)A12SJ(Q , d 2)

(13)

The theoretical curves obtained assuming this model eqs 11, 12, and 13 correspond well to the experimental dependencies of EISF(Q) and QISF(Q), as shown in Figure 5b,c. The fit parameters are given in Table 1.

(8)

16583


Article


Table 1. Parameters Obtained from the Fit of Eqs 11−13 to Experimental EISF and QISFs Assuming a Specific Model of Motion rigid temp (K)

type of spectrometer

210

IN5 IN16 IN5 IN5

250 300

model of motion 2SJ 2SJ + 2SJ 2SJ + 2SJ

fast motion

N0 (%)

N1 (%)

d1 (Ĺ)

± ± ± ±

10.0 ± 0.1

2.54 ± 0.03

18.0 ± 2.0 35.0 ± 1.0

2.52 ± 0.04 2.46 ± 0.04

90.0 64.9 44.4 15.2

0.1 0.5 0.3 0.5

slow motion N2 = 100 − N0 − N1 (%)

d2 [Ĺ]

35.1 ± 0.5 38.0 ± 2.0 50.0 ± 1.0

2.92 ± 0.05 3.18 ± 0.03 4.33 ± 0.03

have that about 15% of the protons do not take part in any molecular motion (at least in the time scale of the neutron scattering experiments). Thermal evolution of N0 means that with increase of temperature more number of water molecules can perform jumps (in the time window of spectrometers), while increase jump length (d2) with temperature causes that water molecules have more energy and they are able to jump between the more distant positions. The problem of proton dynamics in adsorbed water molecules and surface OH groups has been studied in other systems as well, for example, for MCM-41 and SBA-15 by 1H NMR,14 and for cyclodextrin by QENS.32 The authors of these works have reported that in the systems they studied protons perform jumps in hydrogen bonds over a distance of about 1 Å. Moreover, for cyclodextrin, proton reorientations in OH groups and water molecules have been identified to take place over a mean distance of 1.55 Å. In our case, these kinds of motions have not been found from our QENS analysis, but it seems impossible not to have them as well. As they were expected to be faster than the jumps of the entire water molecule, in order to verify their existence, a very tiny quasielastic component of neutron scattering spectra observed at 150 K was additionally analyzed following the procedure presented previously in this paper. The results have revealed another motion characterized by a Lorentzian function of half-width 170.1 μeV on average but due to the low intensity of the quasielastic broadening signal its geometry as well as characteristic distance was impossible to extract from the experimental data. Correlation Times. To perform a more accurate analysis of the characteristic times of the proton motions in SBA-15, based on the analysis of the widths of the Lorentzian functions, describing the quasielastic broadening of the neuron scattering spectra, at each analyzed temperature the correlation times were calculated from the formula:

Thus, we have that at 300 and 250 K, the experimental dependencies of EISF(Q) and QISF(Q) are best described by a model formed by the sum of two two-site jumps between equivalent positions 2SJ + 2SJ (both motions in a similar time scale), while at 210 K the experimental EISF(Q)’s measured on IN5 and IN16 are described separately by a single two-jump model 2SJ and the time scales corresponding to each set of data are largely different. For each temperature, the parameters characterizing the motions were verified by fitting again the neutron scattering spectra with the theoretical curve S(Q,ω) eq 10 corresponding to the assumed model 2SJ + 2SJ (for 300 and 250 K − IN5) and 2SJ (for 210 K − IN5 and IN16). This direct fitting procedure produced the same set of parameters as those obtained from fitting eqs 11−13 to EISF/QISF. Additionally, other models were tested, for instance, the diffusion on the sphere.25 But, it was not possible to fit the experimental spectra with the models and get reasonable parameters. Assignment of Motions. As the results of extensive QENS analysis have shown, two types of proton motions are present separately. Both can be described theoretically by a model of jumps between two equivalent sites (2SJ), although they differ in the time scale and characteristic jump distance d. The first motion (corresponding to the Lorentzian function of greater width) involves jumps between two positions over the distance of 2.5 Å for the whole range of analyzed temperatures. The second motion (corresponding to the Lorentzian function of smaller width) is well described by jumps between two positions over a distance in the range from 2.9 Å (210 K) to 4.3 Å (300 K). Therefore, we find that the jump distance of the first motion is not influenced by the temperature, while for the second motion the characteristic jump distance changes by 50% within the studied temperature range (of 90 K). As the mean pore size of SBA-15 mesopores is much greater than the geometrical size of a water molecule, it is expected that water adsorption inside the pores takes place in layers.12,33 As mentioned above, in the studied material there are no free SiOH groups, implying that all silanol groups are H-bonded to water molecules. Because there are ∼1.8 water molecules adsorbed on the silica surface per OH group, water molecules can be bound to the surface SiOH groups in different combinations - one water molecule hydrogen-bounded to one silanol or two molecules hydrogen-bounded to the same silanol (1 as donor, another as acceptor). It is also possible that water molecules are bounded together. It should also be underlined that the silica surface is not flat, but irregular and rough. Therefore, distances between silanol groups and between water molecules are not strictly defined. Thus, we assign the first motion to jumps of water molecules between nearest neighboring positions, and the second one (over greater distances) to jumps between further neighbors. Moreover, it is worth noting that, according to QENS results, even at 300 K the number of motionless protons is nonzero (cf. Table 1). We

τc =

2ℏ Γ

(14)

where ℏ is the normalized Planck constant, and Γ is the average half-width of the Lorentzian function. The correlation times calculated for the two types of water motion identified in the system versus reciprocal temperature are presented in Figure 6. The characteristic times of the jumps over the distance of 2.5 Å (first motion) are much shorter than those of the jumps over longer distances (second motion). Moreover, as can be seen from Figure 6, the motion with larger correlation times has a stronger temperature dependence. In general, it can be said that in the system two types of motion are present with much different correlation times. As the motions are not cooperative, their temperature dependencies can be described separately by the Arrhenius equation: 16584


Article


activation energy of this motion found from experimental data is 5.2 ± 0.2 (NMR) and 5.6 ± 1.1 kJ/mol (QENS). The second type of motion is slower, with a correlation time of 25 ps (300 K). It involves jumps of water molecules between more spatially separated positions. Its characteristic distance varies from 2.9 Å (210 K) to 4.3 Å (300 K), indicating a strong temperature dependence for this motion. The activation energy of this motion found from experimental data is 16.1 ± 0.3 kJ/ mol (NMR) and 17.3 ± 0.3 kJ/mol (QENS) Finally, the experimental data allowed us also to suggest that at least one other motion is present in the system, but in general its time scale exceeds the measuring range of IN5 and IN16 spectrometers. It is characterized by a correlation time of 8 ps at 150 K.

■

ASSOCIATED CONTENT

* Supporting Information S

Figures showing DSC, FTIR, and TG measurements for SBA15 which preliminary characterized samples are shown. The width of half-maximum of fitted Lorentzian functions for selected temperatures are also included. The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcc.5b02672.

Figure 6. Correlation times of water motions in SBA-15 determined from QENS data (■ IN5, ▲ IN16) vs reciprocal temperature with the fit made with the Arrhenius eq 15. The characteristic distances of jumps d (Å) were found from the fit made assuming a certain model of motion to the experimental QENS data.

⎛E ⎞ τc = τ0 exp⎜ a ⎟ ⎝ RT ⎠

■

(15)

AUTHOR INFORMATION

Corresponding Author

where τ0 is the pre-exponential factor, Ea is the activation energy of a given motion, and R is the gas constant. Results of the fit and the activation energies are also presented in Figure 6. The activation energies are 5.6 ± 1.1 kJ/ mol for the jumps over the shorter distance and 17.3 ± 0.3 kJ/ mol for the jumps over longer distances. Thus, the activation of the first type of jumps requires almost 3 times lower energy, so they can take place at lower temperatures. Furthermore, the activation energies of the two types of motion determined on the basis of the NMR results are 5.2 ± 0.2 and 16.1 ± 0.3 kJ/mol, respectively, so they are in good agreement with the values obtained from the QENS measurements, indicating that both methods are detecting the same types of motion. Finally, the correlation time calculated for the motion identified on the basis of QENS results at 150 K is also presented in Figure 6. As it has been shown, at this temperature its value is at least 10 times lower than for two other motions described in this paper, which explains the impossibility to investigate this particular type of dynamics (most probably of reorientational character), using the same set of spectrometers.

*E-mail: [email protected]. Tel.: +48-61-8295-220. Fax: +48-61-8295-155. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work has been partially financed by the National Science Centre of Poland, Grant No. 2012/05/B/ST3/03176 and by the Operational Programme Human Capital (POKL 4.1.1). The authors would like to thank Dr. Kacper Drużbicki for the discussion during the preparation of this manuscript.

■

REFERENCES

(1) Morishige, K.; Nobuoka, K. X-ray Diffraction Studies of Freezing and Melting of Water Confined in a Mesoporous Adsorbent (MCM41). J. Chem. Phys. 1997, 107, 6965−6969. (2) Yoshida, K.; Yamaguchi, T.; Kittaka, S.; Bellissent-Funel, M.-C.; Fouquet, P. Thermodynamic, Structural, and Dynamic Properties of Supercooled Water Confined in Mesoporous MCM-41 Studied with Calorimetric, Neutron Diffraction, and Neutron Spin Echo Measurements. J. Chem. Phys. 2008, 129, 054702. (3) Kittaka, S.; Sou, K.; Yamaguchi, T.; Tozaki, K. Thermodynamic and FTIR Studies of Supercooled Water Confined to Exterior and Interior of Mesoporous MCM-41. Phys. Chem. Chem. Phys. 2009, 11, 8538−8543. (4) Kittaka, S.; Ueda, Y.; Fujisaki, F.; Iiyama, T.; Yamaguchi, T. Mechanism of Freezing of Water in Contact with Mesoporous Silicas MCM-41, SBA-15 and SBA-16: Role of Boundary Water of Pore Outlets in Freezing. Phys. Chem. Chem. Phys. 2011, 13, 17222−17233. (5) Hansen, E. W.; Schmidt, R.; Stoecker, M.; Akporiaye, D. SelfDiffusion Coefficient of Water Confined in Mesoporous MCM-41 Materials Determined by 1H Nuclear Magnetic Resonance Spin-Echo Measurements. Microporous Mater. 1995, 5, 143−150. (6) Hansen, E. W.; Schmidt, R.; Stoecker, M.; Akporiaye, D. WaterSaturated Mesoporous MCM-41 Systems Characterized by 1H NMR Spin-Lattice Relaxation Times. J. Phys. Chem. 1995, 99, 4148−4154. (7) Takamuku, T.; Yamagami, M.; Wakita, H.; Masuda, Y.; Yamaguchi, T. Thermal Property, Structure, and Dynamics of Supercooled Water in Porous Silica by Calorimetry, Neutron

IV. CONCLUSIONS This study was undertaken to characterize the dynamics of protons in a mesoporous silica material SBA-15 with low water content (15% of pore filling) which was confirmed by DSC and TG methods. Additionally the FTIR spectrum shows that water molecules are hydrogen bonded to the silica surface. As a result of a thorough NMR and QENS study the possibility of confined water exhibiting some translational diffusion was excluded (in the experimental time scale), while two other localized stochastic motions of water molecules were characterized in detail. The first type of motion was described as fast, with a correlation time of 4 ps (300 K) and a characteristic distance equal to 2.5 Å (in the whole temperature range analyzed). It is supposed to be performed by water molecules that jump between neighboring positions. The 16585


Article

The Journal of Physical Chemistry C Scattering, and NMR Relaxation. J. Phys. Chem. B 1997, 101, 5730− 5739. (8) Akporiaye, D.; Hansen, E. W.; Schmidt, R.; Stocker, M. WaterSaturated Mesoporous MCM-41 Systems Characterized by 1H NMR. J. Phys. Chem. 1994, 98, 1926−1928. (9) Takahara, S.; Nakano, M.; Kittaka, S.; Kuroda, Y.; Mori, T.; Hamano, H.; Yamaguchi, T. Neutron Scattering Study on Dynamics of Water Molecules in MCM-41. J. Phys. Chem. B 1999, 103, 5814−5819. (10) Faraone, A.; Liu, L.; Mou, Ch-Y; Shih, P-Ch; Copley, J. R. D.; Chen, S.-H. Translational and Rotational Dynamics of Water in Mesoporous Silica Materials: MCM-41-S and MCM-48-S. J. Chem. Phys. 2003, 119, 3963−3971. (11) Briman, I. M.; Rábiscoul, D.; Diat, O.; Zanotti, J.-M.; Jollivet, P.; Barboux, P.; Gin, S. Impact of Pore Size and Pore Surface Composition on the Dynamics of Confined Water in Highly Ordered Porous Silica. J. Phys. Chem. C 2012, 116, 7021−7028. (12) Milischuk, A. A.; Ladanyi, B. M. Structure and Dynamics of Water Confined in Silica Nanopores. J. Chem. Phys. 2011, 135, 174709. (13) Bourg, I. C.; Steefel, C. I. Molecular Dynamics Simulations of Water Structure and Diffusion in Silica Nanopores. J. Phys. Chem. C 2012, 116, 11556−11564. (14) Grunberg, B.; Emmler, Th.; Gedat, E.; Shenderovich, I.; Findenegg, G. H.; Limbach, H.-H.; Buntkowsky, G. Hydrogen Bonding of Water Confined in Mesoporous Silica MCM-41 and SBA-15 Studied by 1H Solid-State NMR. Chem.Eur. J. 2004, 10, 5689−5696. (15) Gallo, P.; Rapinesi, M.; Rovere, M. Confined Water in the Low Hydration Regime. J. Chem. Phys. 2002, 117, 369−375. (16) Zhao, D.; Feng, J.; Huo, Q.; Melosh, N.; Fredrickson, G. H.; Chmelka, B. F.; Stucky, G. D. Triblock Copolymer Syntheses of Mesoporous Silica with Periodic 50 to 300 Angstrom Pores. Science 1998, 279, 548−552. (17) Kruk, M.; Jaroniec, M.; Ko, Ch.H; Ryoo, R. Characterization of the Porous Structure of SBA-15. Chem. Mater. 2000, 12, 1961−1968. (18) Schuth, F., Sing, K. S. W., Weitkamp, J., Eds. Handbook of Porous Solids; Wiley-VCH: Weinheim, Germany, 2002. (19) https://www.ill.eu/instruments-support/computing-forscience/cs-software/all-software/lamp/. Date of access: May 2014. (20) Colilla, M.; Martinez-Carmona, M.; Sanchez-Salcedo, S.; RuizGonzalez, M. L.; Gonzalez-Calbet, J. M.; Vallet-Regi, M. A Novel Zwitterionic Bioceramic with Dual Antibacterial Capability. J. Mater. Chem. B 2014, 2, 5639−5651. (21) Abadleh, H. A.; Grassian, V. H. FT-IR Study of Water Adsorption on Aluminum Oxide Surfaces. Langmuir 2003, 19, 341− 347. (22) Zhao, X. S.; Lu, G. Q.; Whittaker, A. K.; Millar, G. J.; Zhu, H. Y. Comprehensive Study of Surface Chemistry of MCM-41 Using 29Si CP/MAS NMR, FTIR, Pyridine-TPD, and TGA. J. Phys. Chem. B 1997, 101, 6525−6531. (23) Zhuravlev, L. T. The Surface Chemistry of Amorphous Silica. Zhuravlev Model. Colloids Surf., A 2000, 173, 1−38. (24) Pajzderska, A.; Gonzalez, M. A.; Mielcarek, J.; Wasicki, J. Water Behavior in MCM-41 as a Function of Pore Filling and Temperature Studied by NMR and Molecular Dynamics Simulations. J. Phys. Chem. C 2014, 118, 23701−23710. (25) Bee, M. Quasielastic Neutron Scattering: Principles and Applications in Solid State Chemistry, Biology, and Material Science; Adam Hilger: Bristol and Philadelphia, 1988. (26) Bellissent-Funel, M.-C.; Chen, S. H.; Zanotti, J.-M. SingleParticle Dynamics of Water Molecules in Confined Space. Phys. Rev. E: Stat. Phys., Plasmas, Fluids, Relat. Interdiscip. Top. 1995, 51, 4558− 4569. (27) Zanotti, J.-M.; Bellissent-Funel, M.-C.; Chen, S.-H. Relaxational Dynamics of Supercooled Water in Porous Glass. Phys. Rev. E: Stat. Phys., Plasmas, Fluids, Relat. Interdiscip. Top. 1999, 59, 3084−3093. (28) Bastos, M.; Castro, V.; Mrevlishvili, G.; Teixeira, J. Hydration of DS-DNA and SS-DNA by Neutron Quasielastic Scattering. Biophys. J. 2004, 86, 3822−3827.

(29) Bellissent-Funel, M.-C.; Teixeira, J.; Bradley, K. F.; Chen, S. H.; Crespi, H. L. Single-Particle Dynamics of Hydration Water in Protein. Phys. B (Amsterdam, Neth.) 1992, 180−181, 740−744. (30) Bellissent-Funel, M.-C.; Teixeira, J.; Bradley, K. F.; Chen, S. H. Dynamics of Hydration Water in Protein. J. Phys. I 1992, 2, 995−1001. (31) Buchsteiner, A.; Lerf, A.; Pieper, J. Water Dynamics in Graphite Oxide Investigated with Neutron Scattering. J. Phys. Chem. B 2006, 110, 22328−22338. (32) Steiner, Th.; Saenger, W.; Lechner, R. E. Dynamics of Orientationally Disordered Hydrogen Bonds and of Water Molecules in a Molecular Cage. Mol. Phys. 1991, 72, 1211−1232. (33) Gallo, P.; Rovere, M.; Chen, S.-H. Dynamic Crossover in Supercooled Confined Water: Understanding Bulk Properties Through Confinement. J. Phys. Chem. Lett. 2010, 1, 729−733.

16586


QENS and NMR Study of Water Dynamics in SBA-15 with a Low Water

Recommend Documents